Advertisement
Research Article| Volume 60, 102750, September 2022

Download started.

Ok

Transfer, persistence and recovery of DNA and mRNA vaginal mucosa markers after intimate and social contact with Bayesian network analysis for activity level reporting

Open AccessPublished:July 21, 2022DOI:https://doi.org/10.1016/j.fsigen.2022.102750
Transfer, persistence and recovery of DNA and mRNA vaginal mucosa markers after intimate and social contact with Bayesian network analysis for activity level reporting
Previous ArticleEstimating the time since deposition (TsD) in saliva stains using temporal changes in microbial markers
Next ArticleAssessing the utility of quantitative and qualitative metrics in the DNA quantification process of skeletal remains for autosomal and Y-chromosome STR amplification purposes

      Highlights

      • ●
        Transfer and persistence data of DNA and mRNA vaginal mucosa marker
      • ●
        Bayesian network analysis to evaluate the evidence at activity level
      • ●
        The LR value was mainly affected by the high DNA quantity

      Abstract

      In sexual assault cases, it can be challenging to identify the type of body fluids/ cell types present in a crime scene sample, especially the origin of epithelial cells. Therefore, more labs are applying mRNA body fluid analysis for saliva, skin and vaginal mucosa markers. To address activity level propositions, it is necessary to assign probabilities of transfer, persistence, prevalence and recovery of DNA and mRNA markers. In this study we analysed 158 samples (fingernail swabs, penile swabs and boxershorts) from 12 couples collected at different time points post intimate contact and after non-intimate contact in order to detect DNA from the person of interest (POI) and mRNA vaginal mucosa markers. Samples were DNA and RNA co-extracted and analysed with PowerPlex®Fusion 6C System and 19-plex mRNA primer mix respectively, using Endpoint PCR and the CE platform. Vaginal mucosa was detected up to 36 h post intimate contact, but also detected in one non-intimate contact sample. In 94% of intimate contact and 50 % of non-intimate contact samples the DNA results support the proposition that POI is the donor (LR ≥ 10,000). There was a strong association between the detection of vaginal mucosa and the average RFU value of the POI. The data were used to instantiate a comprehensive Bayesian network to evaluate the evidence at activity level, given alternate propositions conditioned upon indirect or direct transfer events. It is shown that the value of the evidence is mainly affected by the high DNA quantity (measured as mean RFU) that is recovered from the POI. The detection of vaginal mucosa had low impact upon the resultant likelihood ratio.

      Keywords

      1. Introduction

      In sexual assault cases, the source of the body fluid and DNA is often in question, especially if the DNA profile originated from epithelial cells (skin or vaginal mucosa). For instance, if DNA from the victim is detected on a penile swab recovered from the suspect, the evaluation of the cell-type may assist the investigation. If the prosecution propose that a sexual assault was committed by the suspect, and DNA from the victim was transferred as a result, the defence might claim that only social contact occurred i.e., the DNA originated from epithelial skin cells from the victim and not vaginal epithelial cells. Hence, a method to detect epithelial cells derived from vaginal mucosa might add value to the evaluation of the evidence.
      Methods for body fluid identification that are commonly used by forensic laboratories, except for the detection of spermatozoa, are presumptive tests that only provide an indication that the body fluid of interest may be present. Tests are mainly conducted for the presence of blood, seminal fluid and saliva, and are often based on a chemical colour reaction or an immunology antibody-antigen interaction [
      • Harbison S.A.
      • Fleming R.
      Forensic body fluid identification: state of the art.
      Res. Rep. Forensic Med. Sci. 2016; 6: 11-23
      ,
      • Virkler K.
      • Lednev I.K.
      Analysis of body fluids for forensic purposes: from laboratory testing to non-destructive rapid confirmatory identification at a crime scene.
      Forensic Sci. Int. 2009; 188: 1-17
      ]. These tests are easy and quick to perform, but they are not always accurate and can give false results. Because of these shortcomings, alternative tests that have higher specificity and simultaneously test a whole range of body fluids are preferable. Since Juusola and Ballantyne [
      • Juusola J.
      • Ballantyne J.
      Messenger RNA profiling: a prototype method to supplant conventional methods for body fluid identification.
      Forensic Sci. Int. 2003; 135: 85-96
      ] introduced a method to measure gene expression of mRNA markers, a selection has been successfully combined in a multiplex [
      • Haas C.
      • Hanson E.
      • Anjos M.J.
      • Bar W.
      • Banemann R.
      • Berti A.
      • Borges E.
      • Bouakaze C.
      • Carracedo A.
      • Carvalho M.
      • Castella V.
      • Choma A.
      • De Cock G.
      • Dotsch M.
      • Hoff-Olsen P.
      • Johansen P.
      • Kohlmeier F.
      • Lindenbergh P.A.
      • Ludes B.
      • Maronas O.
      • Moore D.
      • Morerod M.L.
      • Morling N.
      • Niederstatter H.
      • Noel F.
      • Parson W.
      • Patel G.
      • Popielarz C.
      • Salata E.
      • Schneider P.M.
      • Sijen T.
      • Sviezena B.
      • Turanska M.
      • Zatkalikova L.
      • Ballantyne J.
      RNA/DNA co-analysis from blood stains--results of a second collaborative EDNAP exercise.
      Forensic Sci. Int. Genet. 2012; 6: 70-80
      ,
      • Haas C.
      • Hanson E.
      • Anjos M.J.
      • Banemann R.
      • Berti A.
      • Borges E.
      • Carracedo A.
      • Carvalho M.
      • Courts C.
      • De Cock G.
      • Dotsch M.
      • Flynn S.
      • Gomes I.
      • Hollard C.
      • Hjort B.
      • Hoff-Olsen P.
      • Hribikova K.
      • Lindenbergh A.
      • Ludes B.
      • Maronas O.
      • McCallum N.
      • Moore D.
      • Morling N.
      • Niederstatter H.
      • Noel F.
      • Parson W.
      • Popielarz C.
      • Rapone C.
      • Roeder A.D.
      • Ruiz Y.
      • Sauer E.
      • Schneider P.M.
      • Sijen T.
      • Court D.S.
      • Sviezena B.
      • Turanska M.
      • Vidaki A.
      • Zatkalikova L.
      • Ballantyne J.
      RNA/DNA co-analysis from human saliva and semen stains--results of a third collaborative EDNAP exercise.
      Forensic Sci. Int. Genet. 2013; 7: 230-239
      ,
      • Haas C.
      • Hanson E.
      • Anjos M.J.
      • Ballantyne K.N.
      • Banemann R.
      • Bhoelai B.
      • Borges E.
      • Carvalho M.
      • Courts C.
      • De Cock G.
      • Drobnic K.
      • Dotsch M.
      • Fleming R.
      • Franchi C.
      • Gomes I.
      • Hadzic G.
      • Harbison S.A.
      • Harteveld J.
      • Hjort B.
      • Hollard C.
      • Hoff-Olsen P.
      • Huls C.
      • Keyser C.
      • Maronas O.
      • McCallum N.
      • Moore D.
      • Morling N.
      • Niederstatter H.
      • Noel F.
      • Parson W.
      • Phillips C.
      • Popielarz C.
      • Roeder A.D.
      • Salvaderi L.
      • Sauer E.
      • Schneider P.M.
      • Shanthan G.
      • Court D.S.
      • Turanska M.
      • van Oorschot R.A.
      • Vennemann M.
      • Vidaki A.
      • Zatkalikova L.
      • Ballantyne J.
      RNA/DNA co-analysis from human menstrual blood and vaginal secretion stains: results of a fourth and fifth collaborative EDNAP exercise.
      Forensic Sci. Int. Genet. 2014; 8: 203-212
      ,
      • van den Berge M.
      • Carracedo A.
      • Gomes I.
      • Graham E.A.
      • Haas C.
      • Hjort B.
      • Hoff-Olsen P.
      • Maronas O.
      • Mevag B.
      • Morling N.
      • Niederstatter H.
      • Parson W.
      • Schneider P.M.
      • Court D.S.
      • Vidaki A.
      • Sijen T.
      A collaborative European exercise on mRNA-based body fluid/skin typing and interpretation of DNA and RNA results.
      Forensic Sci. Int. Genet. 2014; 10: 40-48
      ,
      • Lindenbergh A.
      • de Pagter M.
      • Ramdayal G.
      • Visser M.
      • Zubakov D.
      • Kayser M.
      • Sijen T.
      A multiplex (m)RNA-profiling system for the forensic identification of body fluids and contact traces.
      Forensic Sci. Int. Genet. 2012; 6: 565-577
      ,
      • Albani P.P.
      • Fleming R.
      Developmental validation of an enhanced mRNA-based multiplex system for body fluid and cell type identification.
      Sci. Justice. 2019; 59: 217-227
      ].
      The most common mRNA vaginal mucosa markers that have been identified and tested for forensic application are Mucin 4 (MUC4), Human beta-defensin 1 (HBD1), Myozenin 1 (MYOZ1), and Cytochrome P450, Family 2 Subfamily B Polypeptide 7 Pseudogene 1 (CYP2B7P1) [
      • Lindenbergh A.
      • de Pagter M.
      • Ramdayal G.
      • Visser M.
      • Zubakov D.
      • Kayser M.
      • Sijen T.
      A multiplex (m)RNA-profiling system for the forensic identification of body fluids and contact traces.
      Forensic Sci. Int. Genet. 2012; 6: 565-577
      ,
      • Albani P.P.
      • Fleming R.
      Developmental validation of an enhanced mRNA-based multiplex system for body fluid and cell type identification.
      Sci. Justice. 2019; 59: 217-227
      ,
      • Jakubowska J.
      • Maciejewska A.
      • Pawlowski R.
      • Bielawski K.P.
      mRNA profiling for vaginal fluid and menstrual blood identification.
      Forensic Sci. Int. Genet. 2013; 7: 272-278
      ,
      • Roeder A.D.
      • Haas C.
      mRNA profiling using a minimum of five mRNA markers per body fluid and a novel scoring method for body fluid identification.
      Int. J. Leg. Med. 2013; 127: 707-721
      ,
      • Hanson E.K.
      • Ballantyne J.
      Highly specific mRNA biomarkers for the identification of vaginal secretions in sexual assault investigations.
      Sci. Justice. 2013; 53: 14-22
      ,
      • Richard M.L.
      • Harper K.A.
      • Craig R.L.
      • Onorato A.J.
      • Robertson J.M.
      • Donfack J.
      Evaluation of mRNA marker specificity for the identification of five human body fluids by capillary electrophoresis.
      Forensic Sci. Int. Genet. 2012; 6: 452-460
      ,
      • Cossu C.
      • Germann U.
      • Kratzer A.
      • Bär W.
      • Haas C.
      How specific are the vaginal secretion mRNA-markers HBD1 and MUC4?.
      Forensic Sci. Int. Genet. Suppl. Ser. 2009; 2: 536-537
      ,
      • Blackman S.
      • Stafford-Allen B.
      • Hanson E.K.
      • Panasiuk M.
      • Brooker A.-L.
      • Rendell P.
      • Ballantyne J.
      • Wells S.
      Developmental validation of the ParaDNA® body fluid ID system – a rapid multiplex mRNA-profiling system for the forensic identification of body fluids.
      Forensic Sci. Int. Genet. 2018; 37: 151-161
      ,
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ]. The MUC4 and the HBD1 markers have shown to be less specific as they often cross-react with other body fluids, especially saliva/buccal swabs and nasal mucosa [
      • Haas C.
      • Hanson E.
      • Anjos M.J.
      • Ballantyne K.N.
      • Banemann R.
      • Bhoelai B.
      • Borges E.
      • Carvalho M.
      • Courts C.
      • De Cock G.
      • Drobnic K.
      • Dotsch M.
      • Fleming R.
      • Franchi C.
      • Gomes I.
      • Hadzic G.
      • Harbison S.A.
      • Harteveld J.
      • Hjort B.
      • Hollard C.
      • Hoff-Olsen P.
      • Huls C.
      • Keyser C.
      • Maronas O.
      • McCallum N.
      • Moore D.
      • Morling N.
      • Niederstatter H.
      • Noel F.
      • Parson W.
      • Phillips C.
      • Popielarz C.
      • Roeder A.D.
      • Salvaderi L.
      • Sauer E.
      • Schneider P.M.
      • Shanthan G.
      • Court D.S.
      • Turanska M.
      • van Oorschot R.A.
      • Vennemann M.
      • Vidaki A.
      • Zatkalikova L.
      • Ballantyne J.
      RNA/DNA co-analysis from human menstrual blood and vaginal secretion stains: results of a fourth and fifth collaborative EDNAP exercise.
      Forensic Sci. Int. Genet. 2014; 8: 203-212
      ,
      • Jakubowska J.
      • Maciejewska A.
      • Pawlowski R.
      • Bielawski K.P.
      mRNA profiling for vaginal fluid and menstrual blood identification.
      Forensic Sci. Int. Genet. 2013; 7: 272-278
      ,
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ], but the two remaining markers, MYOZ1 and CYP2B7P1, have proved to be more promising markers for the detection of vaginal mucosa [
      • Haas C.
      • Hanson E.
      • Anjos M.J.
      • Ballantyne K.N.
      • Banemann R.
      • Bhoelai B.
      • Borges E.
      • Carvalho M.
      • Courts C.
      • De Cock G.
      • Drobnic K.
      • Dotsch M.
      • Fleming R.
      • Franchi C.
      • Gomes I.
      • Hadzic G.
      • Harbison S.A.
      • Harteveld J.
      • Hjort B.
      • Hollard C.
      • Hoff-Olsen P.
      • Huls C.
      • Keyser C.
      • Maronas O.
      • McCallum N.
      • Moore D.
      • Morling N.
      • Niederstatter H.
      • Noel F.
      • Parson W.
      • Phillips C.
      • Popielarz C.
      • Roeder A.D.
      • Salvaderi L.
      • Sauer E.
      • Schneider P.M.
      • Shanthan G.
      • Court D.S.
      • Turanska M.
      • van Oorschot R.A.
      • Vennemann M.
      • Vidaki A.
      • Zatkalikova L.
      • Ballantyne J.
      RNA/DNA co-analysis from human menstrual blood and vaginal secretion stains: results of a fourth and fifth collaborative EDNAP exercise.
      Forensic Sci. Int. Genet. 2014; 8: 203-212
      ,
      • Hanson E.K.
      • Ballantyne J.
      Highly specific mRNA biomarkers for the identification of vaginal secretions in sexual assault investigations.
      Sci. Justice. 2013; 53: 14-22
      ].
      There are numerous studies published on transfer, persistence, prevalence and recovery (TPPR) of DNA in samples collected from fingernails [
      • van den Berge M.
      • Ozcanhan G.
      • Zijlstra S.
      • Lindenbergh A.
      • Sijen T.
      Prevalence of human cell material: DNA and RNA profiling of public and private objects and after activity scenarios.
      Forensic Sci. Int. Genet. 2016; 21: 81-89
      ,
      • Lacerenza D.
      • Aneli S.
      • Omedei M.
      • Gino S.
      • Pasino S.
      • Berchialla P.
      • Robino C.
      A molecular exploration of human DNA/RNA co-extracted from the palmar surface of the hands and fingers.
      Forensic Sci. Int. Genet. 2016; 22: 44-53
      ,
      • Cook O.
      • Dixon L.
      The prevalence of mixed DNA profiles in fingernail samples taken from individuals in the general population.
      Forensic Sci. Int. Genet. 2007; 1: 62-68
      ,
      • Malsom S.
      • Flanagan N.
      • McAlister C.
      • Dixon L.
      The prevalence of mixed DNA profiles in fingernail samples taken from couples who co-habit using autosomal and Y-STRs.
      Forensic Sci. Int. Genet. 2009; 3: 57-62
      ,
      • Matte M.
      • Williams L.
      • Frappier R.
      • Newman J.
      Prevalence and persistence of foreign DNA beneath fingernails.
      Forensic Sci. Int. Genet. 2012; 6: 236-243
      ] and penile swabs [
      • Fonneløp A.E.
      • Johannessen H.
      • Heen G.
      • Molland K.
      • Gill P.
      A retrospective study on the transfer, persistence and recovery of sperm and epithelial cells in samples collected in sexual assault casework, Forensic Science.
      Forensic Sci. Int. Genet. 2019; 43
      ,
      • Bouzga M.M.
      • Dørum G.
      • Gundersen K.
      • Kohler P.
      • Hoff-Olsen P.
      • Fonneløp A.E.
      Is it possible to predict the origin of epithelial cells? – a comparison of secondary transfer of skin epithelial cells versus vaginal mucous membrane cells by direct contact.
      Sci. Justice. 2020; 60: 234-242
      ,
      • Cina S.J.
      • Collins K.A.
      • Pettenati M.J.
      • Fitts M.
      Isolation and identification of female DNA on postcoital penile swabs.
      Am. J. Forensic Med. Pathol. 2000; 21: 97-100
      ,
      • Farmen R.K.
      • Haukeli I.
      • Ruoff P.
      • Froyland E.S.
      Assessing the presence of female DNA on post-coital penile swabs: relevance to the investigation of sexual assault.
      J. Forensic Leg. Med. 2012; 19: 386-389
      ,
      • Kaarstad K.
      • Rohde M.
      • Larsen J.
      • Eriksen B.
      • Thomsen J.L.
      The detection of female DNA from the penis in sexual assault cases.
      J. Forensic Leg. Med. 2007; 14: 159-160
      ]. However there are limited equivalent studies that have been published on TPPR of mRNA vaginal secretion markers. Hanson et al. [
      • Hanson E.K.
      • Ballantyne J.
      Highly specific mRNA biomarkers for the identification of vaginal secretions in sexual assault investigations.
      Sci. Justice. 2013; 53: 14-22
      ] successfully detected the markers MYOZ1 and CYP2B7P1 in a penile swab, boxershorts and a finger swab collected post intercourse and digital penetration, while a penile swab taken prior to the intercourse were negative for these two markers. Blackman et al. [
      • Blackman S.
      • Stafford-Allen B.
      • Hanson E.K.
      • Panasiuk M.
      • Brooker A.-L.
      • Rendell P.
      • Ballantyne J.
      • Wells S.
      Developmental validation of the ParaDNA® body fluid ID system – a rapid multiplex mRNA-profiling system for the forensic identification of body fluids.
      Forensic Sci. Int. Genet. 2018; 37: 151-161
      ] reported the detection of marker CYP2B7P1 in a penile swab collected 36 h after intercourse. However, the latter two studies include very few participants and samples. Other studies have investigated the prevalence of mRNA body fluid markers collected from fingernail, palms [
      • van den Berge M.
      • Ozcanhan G.
      • Zijlstra S.
      • Lindenbergh A.
      • Sijen T.
      Prevalence of human cell material: DNA and RNA profiling of public and private objects and after activity scenarios.
      Forensic Sci. Int. Genet. 2016; 21: 81-89
      ,
      • Lacerenza D.
      • Aneli S.
      • Omedei M.
      • Gino S.
      • Pasino S.
      • Berchialla P.
      • Robino C.
      A molecular exploration of human DNA/RNA co-extracted from the palmar surface of the hands and fingers.
      Forensic Sci. Int. Genet. 2016; 22: 44-53
      ] and penile swabs [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ]. In the latter study, van den Berge et al. [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ] detected the vaginal mucosa marker MUC4 in 46 % of the penile swab samples.
      Since more labs are implementing mRNA body fluid analysis in sexual assault cases, there is a need to improve knowledge of TPPR of mRNA vaginal mucosa markers from samples collected. In this paper we study transfer and persistence of mRNA vaginal mucosa markers (MUC4, MYOZ1 and CYP2B7P1) collected from boxershorts, fingernail and penile swabs by 12 couples after intimate contact at different time points ranging from 0 to 36 h. In addition, we present data on the same sample types when no intimate (only social) contact has occurred. The dataset also includes the DNA results from the samples (co-extraction of RNA and DNA). The overall aims of this study were to investigate: (1) How long after intimate contact is it possible to detect mRNA vaginal mucosa markers in fingernail or penile swabs; (2) If mRNA vaginal mucosa markers can be detected in fingernail swabs, penile swabs or boxershorts if only social contact has occurred; and (3) The association between the mRNA and DNA results.
      The hierarchy of propositions framework is used to evaluate evidence at different levels [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ]. The sub-source level directly addresses the donor of a DNA profile, whereas the source level addresses the type of body fluid that is present in the crime-stain. Activity level propositions are required to address the case circumstances and require a formal assessment of indirect and direct transfer, persistence, and recovery of the DNA profile and mRNA from known individuals and background from unknown individuals. The data collected in this study were used in Bayesian networks to evaluate the evidence using activity level propositions to generate likelihood ratios [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ].

      2. Method

      2.1 Ethical declaration

      This study was approved by the Data protection officer (DPO) at Oslo University Hospital prior to initiating this project (reference 20/13115). The project was carried out according to the approved procedures and protocols, and all participants have given informed consent.

      2.2 Sample collection

      Twenty four participants (12 couples), age 23 – 56 years old, in steady relationships volunteered to take part in this study. Information about living arrangement was not collected, but it was assumed that the couples either lived together or spent substantial time in each other’s household. The couples received a sampling kit and instructions on how to collect the samples. DNA reference samples were collected from both female and male participants, and all samples were collected from the male participants themselves. Fingernail and penile swabs were collected at five different time points post intimate contact, and with no prior intimate contact (further described in 2.2.3, 2.2.4). In addition, a pair of boxershorts worn by the male after intimate contact and with no prior intimate contact was collected from each couple. Overall, fourteen experimental samples were obtained. One couple did not submit the 12, 18 and 36 h fingernail samples. Another couple did not submit the boxershorts post intimate contact, and the collection of the 18, 24 and 36 h samples (both fingernail and penile swabs) were not carried out according to instructions, hence these samples were excluded from further analysis. The data set includes a total of 122 intimate contact and 36 non-intimate contact samples (both DNA and RNA profiles).

      2.2.1 Preparation of sample kits

      Kits with all equipment needed to collect samples, were supplied to participants: cotton swabs (mwe, Tubed Sterile Dryswab™, MW1041), water ampules (Miwana, 2 mL, 0.9 % NaCl), pre-labelled paper bags for swabs, boxershorts packed in pre-labelled paper bags and a guidance with instruction on how and when to collect the samples.
      The boxershorts were newly bought, in a multipack, and consisted of 95% cotton and 5% elastane (black and white coloured) or 85 % cotton, 10 % viscose and 5 % elastane (grey coloured), sized L and XL. The boxershorts were cleaned in a washing machine (approx. 2 h washing programme at 60 °C, with detergent) and UV-irradiated for 30 min on each side prior to the experiment to remove any DNA and RNA present. A total of six negative control samples were collected from six random boxershorts (one of each colour and size) to control for any possible level of background DNA and RNA after cleaning and UV-irradiation. The samples were DNA and RNA co-extracted and analysed as explained in Section 2.3. While all six DNA fractions were analysed, only three RNA extractions with the highest RNA quantification values (0.032–0.034 ng/ul) underwent RNA profiling with 0.5, 1 and 3 µL cDNA input. All samples analysed were negative, i.e. no STR alleles or mRNA housekeeping genes detected.

      2.2.2 Pilot study

      To determine the optimal time points for sampling after intimate contact, a pilot study was carried out with three couples. Fingernail and penile swab samples were collected at 0, 12, 24, 48 and 72 h post intimate contact. No mRNA vaginal secretion markers (based on MYOZ1 and CYP2B7P1 markers) were detected in any of the pilot samples collected beyond 12 h. Based on these results the time points for the main experiment were set to 0, 12, 18, 24 and 36 h after intimate contact.

      2.2.3 Sample collection after intimate contact (transfer and persistence samples)

      “Intimate contact” was defined as vaginal penetration by penis and by fingers from dominant hand. Swabs from underneath all fingernails of dominant hand and penile swabs (shaft) were collected with a moistened cotton swab, applying medium pressure at 0, 12, 18, 24 and 36 h post intimate contact. A new intimate contact was required for each time point. In addition, a pair of boxershorts worn for 5–6 h after intimate contact was collected. The couples were instructed to refrain from showering between intimate contact and sampling, but hand wash was permitted. No restrictions with regards to social contact were given. The participants were asked to note the approximate number of hand washes between intimate contact and sampling on the pre-labelled paper bags.

      2.2.4 Sample collection after non-intimate contact (prevalence and background samples) and DNA reference sample

      To monitor the level of mRNA and DNA where no intimate contact has occurred, a fingernail swab, a penile swab and a pair of boxershorts (worn for 5–6 h) were collected from each couple as described in Section 2.2.3, but with no prior intimate contact (at least 72 h since last intimate contact). In this paper we define prevalent DNA from known sources and background DNA from unknown sources, as described by Gill et al. [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ].
      In addition a mouth swab was collected as a DNA reference sample from each of the 24 participants.

      2.3 Sample processing

      The tips of the cotton swabs were cut off and placed in a 1.5 mL extraction tube. The crotch area (approx. 18 × 20 cm) of inside boxershorts was sampled by tape-lifting using a mini-tape (Scenesafe FAST™, K545). The adhesive part of the mini-tape was cut into smaller pieces and placed into the extraction tube. The workflow of the experiment samples (intimate and non-intimate contact samples) is presented in Fig. 1.
      Fig. 1
      Fig. 1An overview of the workflow of the processes for the experiment samples.

      2.3.1 DNA and RNA co-extraction

      All experimental samples were co-extracted using QIAamp DNA mini kit (QIAGEN) and mirVANA™ miRNA isolation kit (invitrogen by Thermo Fisher Scientific). The applied method was retrieved from the paper by Lindenbergh et al. [
      • Lindenbergh A.
      • de Pagter M.
      • Ramdayal G.
      • Visser M.
      • Zubakov D.
      • Kayser M.
      • Sijen T.
      A multiplex (m)RNA-profiling system for the forensic identification of body fluids and contact traces.
      Forensic Sci. Int. Genet. 2012; 6: 565-577
      ] with a few exceptions described by Wang et al. [
      • Wang S.
      • Shanthan G.
      • Bouzga M.M.
      • Dinh H.M.T.
      • Haas C.
      • Fonneløp A.E.
      Evaluating the performance of five up-to-date DNA/RNA coextraction methods for forensic application.
      Forensic Sci. Int. 2021; 328
      ]. A negative extraction control was included in each co-extraction run; 23 controls were analysed in total. One STR allele was detected at one locus in two controls (109 and 121 RFU respectively), and the mRNA housekeeping genes were detected in a third control at 5 µL cDNA input (481 and 110 RFU). The detected alleles were at low level and the negative controls were considered to be clear.

      2.3.2 DNA extraction of reference samples

      The DNA reference samples were extracted using the BioRobot EZ1 with EZ1 DNA Investigator Kit and EZ1 DNA Investigator Card (QIAGEN). For pre-treatment of samples 190 µL G2 buffer and 10 µL Proteinase K was added followed by incubation at 56 °C at 600 rpm for at least 1.5 h. The extraction was performed according to the manufacture’s protocol using the “Trace protocol”. The elution volume was set to 200 µL.

      2.3.3 RNA analysis

      The DNAse treatment was performed on RNA extracts with TURBO DNA-free™ kit (invitrogen by Thermo Fisher Scientific) according to the manufacturer’s protocol, using 2 µL (4 units) of TURBO DNAse. The RNA extracts were quantified using Quantifluor® RNA System on the Quantus® Fluorometer (Promega) according to the manufactures recommendations, using low standard calibration. Five µL of extract was added to 200 µL of QuantiFluor® Dye working solution. If the concentration of sample was higher than standard, the samples were diluted by adding 1 µL to the working solution to ensure it was within the linear scale of the standard. If the samples were still higher than standard after dilution, the samples were measured at high standard calibration.
      The complementary DNA (cDNA) synthesis was performed using SuperScript® IV Reverse Transcriptase (invitrogen by Thermo Fisher Scientific) according to the manufacturer’s protocol, using 20 µL reverse transcription reaction volume. Ten µL RNA extracts was added to the reaction mix containing the SuperScript® IV Reverse Transcriptase enzyme (RT positive reactions). If the PCR failed due to possible overload of RNA in RT reactions or presence of inhibitors, a diluted RT reaction was performed using 2 or 5 µL RNA extract. This was the case for seven samples. Negative RT controls (2 µL RNA extract, 8 µL DEPC-treated water (Ambion™, Invitrogen by Thermo Fisher Scientific) and reaction mix without enzyme) were included for each experiment sample to detect any genomic DNA still present in the RNA extract. Housekeeping genes were not detected in any of the RT negative controls, and the samples were clear.
      For mRNA profiling the QIAGEN Multiplex PCR Kit (QIAGEN) and a RNA 19-plex optimized by the NFI (personal communication) was used (Supplementary material 1, Table S1). The markers included in the 19-plex are described by van den Berge et al. [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ] (“Cell-typer V3” multiplex), but the two skin markers are replaced by two gender specific markers (XIST and RPS4Y1) [
      • van den Berge M.
      • Sijen T.
      A male and female RNA marker to infer sex in forensic analysis.
      Forensic Sci. Int. Genet. 2017; 26: 70-76
      ], and with optimized primer sequences [
      • van den Berge M.
      • Sijen T.
      Extended specificity studies of mRNA assays used to infer human organ tissues and body fluids.
      Electrophoresis. 2017; 38: 3155-3160
      ]. The markers included were: blood markers HBB, ALAS2, CD93; saliva markers HTN3 and STATH; semen markers SEMG1, PRM1 and KLK3; vaginal mucosa markers MUC4, MYOZ1 and CYP2B7P1; menstrual secretion markers MMP7, MMP10 and MMP11; nasal mucosa markers BPIFA1; male marker RPS4Y1; female marker XIST; and housekeeping genes ACTB and 18S-rRNA. The multiplex PCR reaction consisted of 12.5 µL 2x QIAGEN Multiplex PCR Master Mix, 5 µL 5x primer mix, 1–7 µL cDNA and 0.5–6.5 µL RNase-free water resulting in 25 µL reaction volume. This multiplex was chosen as it is the validated method in our laboratory.
      For each sample three parallels of cDNA were analysed using 1, 3 and 5 µL cDNA input. Samples in the pilot study were amplified with 0.5, 1 and 3 µL cDNA input; all positive samples, i.e., those where mRNA vaginal markers were detected, were additionally run with 5 µL cDNA input. In addition, samples that showed an increasing mRNA pattern, i.e., mRNA peaks for marker MYOZ1 or CYP2B7P1 detected on only the higher input amounts were amplified with 7 µL cDNA to confirm the presence of these markers. RT negative samples were amplified with 3 µL cDNA input (1 µL for pilot samples).
      Amplification was carried out using a Veriti® 96-Well Thermal Cycler (Applied Biosystems®) using the following settings: 95 °C for 15 min, 33 cycles of 94 °C 20 s, 64 °C 30 s, 72 °C 40 s and a final extension at 60 °C for 45 min
      The PCR products of RT positive samples were purified with Performa DTR Gel filtration cartridge (EdgeBio by BioCat) prior to electrophoresis. The cartridge tubes were prepared by repeated centrifugation at 14,000 rpm for 3 and 2 min respectively; liquid was brought back to cartridge before the second centrifugation. The cartridges were placed into cleaned 1.5 mL tubes. Ten µL of PCR product was added to the cartridge and the purified PCR products were eluted by centrifugation at 9600 rpm for 2 min
      One µL of the PCR-products were mixed with 8.9 µL Hi-Di Formamide and 0.1 µL GeneScan™ 500 LIZ™ dye Size Standard (Applied Biosystems™), and injected onto the 3500xl Genetic Analyzer (Applied Biosystems™) at 1.2 kV for 10 s, using POP-4 polymer. The results were analysed using the GeneMapper® ID-X Software version 1.6 (Applied Biosystems™) and the limit of detection (LOD) for the alleles was set to 50 RFU.
      For the detection of body fluids, scoring categories described in Lindenbergh et al. [
      • Lindenbergh A.
      • Maaskant P.
      • Sijen T.
      Implementation of RNA profiling in forensic casework.
      Forensic Sci. Int. Genet. 2013; 7: 159-166
      ] were followed. A valid RNA profile was defined as a profile where both housekeeping genes were detected. The following scoring categories were applied:
      • –
        Body fluid detected: ≥ 50 % of the body fluid markers were observed (based on the number of valid RNA profiles and the number of markers detected for the actual body fluid)
      • –
        Body fluid sporadically detected: < 50 % of body fluid markers were observed
      • –
        Body fluid not detected: no mRNA markers for the body fluid observed
      • –
        Observed and fits with: If a marker is detected and co-expressed with a different body fluid, for instance the presence of saliva marker STATH in nasal mucosa.

      2.3.4 DNA analysis

      DNA fractions and reference samples were quantified with PowerQuant® System (Promega) and amplified with Promega’s PowerPlex®Fusion 6C System (25 µL reaction volume, 1 ng DNA input, 29 amplification cycles) as recommended by the manufacturer. Amplification was carried out using a Veriti® 96-Well Thermal Cycler (Applied Biosystems™). Samples were injected onto the Applied Biosystems 3500xl Genetic Analyzer at 1.2 kV for 24 s. Results were analysed using the GeneMapper® ID-X Software version 1.6 (Applied Biosystems™) and the analytical threshold (AT) for the alleles was set to 100 RFU. No stochastic threshold was applied. DNA profiles were compared to reference profiles from participants (donor and partner), and the number of observed alleles, loci, total RFU value of the profile and mixture proportion of each contributor were recorded. Homozygote alleles were counted as two. For the data analysis the amelogenin marker and the three Y-chromosome STRs were disregarded. Unknown contributors (background DNA) were assigned if the alleles did not match the reference profiles of donor or the female partner (person of interest).
      A DNA mixture was assigned if there were three or more alleles at a locus. Sub-source likelihood ratios (LR) were calculated using EuroForMix v.3.2.0 [
      • Bleka Ø.
      • Storvik G.
      • Gill P.
      EuroForMix: an open source software based on a continuous model to evaluate STR DNA profiles from a mixture of contributors with artefacts.
      Forensic Sci. Int. Genet. 2015; 21: 35-44
      ] under two alternative propositions:
      Hp: The DNA is from the person of interest (POI).
      Hd: The DNA is from an unknown individual, unrelated to POI.
      The donor was conditioned under both propositions. If background alleles were detected, an unknown contributor was added to both the numerator and denominator. If the value of the evidence supports the prosecution’s proposition with a threshold LR ≥ 10 000 the DNA results were interpreted as a positive finding of 'strong support' for the first proposition [

      ENFSI, E.N.F.S.I. guideline for evaluative reporting in forensic science, 2016.

      ] and these results are listed in 3.1, 3.2, 3.3 (description of the results). The unfiltered dataset, including data for all LR values, is given in the Supplementary material 6.
      For one participant, a tri-allelic vWA marker was observed. Since EFM does not extend to somatic mutations, the locus was disregarded when the LR and mixture proportions were calculated.
      Adjusted RFU values were used as quantitation variables as follows: The total RFUtot of the DNA profile was divided by the number of STR loci (m), i.e. 23. The average RFU (RFU̅) per locus was multiplied by the dilution factor (dl) for the PCR input (based on DNA concentration) and the mixture proportion (MX) of the (POI), as described by Gill et al. [
      • Gill P.
      • Bleka Ø.
      • Fonneløp A.E.
      RFU derived LRs for activity level assignments using Bayesian Networks.
      Forensic Sci. Int. Genet. 2022; 56102608
      ]:
      RFU¯POI=MX×RFUtotm×dl
      (1)


      t-tests were carried out to determine if expected RFU̅POI was the same between intimate contact and non-intimate contact samples (5% significance level).
      Statistical analysis and the making of figures presented in 3.1, 3.2, 3.3, were carried out in STATA 16.1 and R version 4.1.1 (www.r-project.org) using package tidyverse version 1.3.1.

      2.4 Bayesian network (BN) analysis

      In forensic casework, scientists are often asked to evaluate the evidence given activity level propositions. This can be addressed by a Bayesian network, which is a graphical tool to evaluate complex probabilities [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ,
      • Taylor D.
      • Biedermann A.
      • Hicks T.
      • Champod C.
      A template for constructing Bayesian networks in forensic biology cases when considering activity level propositions.
      Forensic Sci. Int. Genet. 2018; 33: 136-146
      ,
      • Biedermann A.
      • Taroni F.
      Bayesian networks for evaluating forensic DNA profiling evidence: a review and guide to literature.
      Forensic Sci. Int. Genet. 2012; 6: 147-157
      ]. In this section, we introduce a BN model and in Section 3.5 we demonstrate a case example using the model informed by the data obtained in this study.

      2.4.1 Case circumstances

      The Bayesian network described can be generalised for a particular set of case circumstances.
      • 1)
        A victim claims to be sexually assaulted by a suspect and alleges that vaginal penetration occurred.
      • 2)
        The victim and the suspect have had previous non-intimate contact. They may co-habit or share facilities in an apartment, for example.
      • 3)
        The suspect denies the allegations stating that he only had social contact with the victim.
      • 4)
        There is no allegation that the assault was committed by an unknown individual
      Intimate contact samples are collected from the suspect within 36 h of the alleged offence taking place. These samples include: (a) penile swabs (b) fingernail swabs and/or (c) boxershorts worn at the time of the alleged incident. Note that BN analyses are carried out per item of evidence. Analysis of DNA and test for vaginal mucosa is carried out concurrently.
      Sub-source likelihood ratios are calculated to support the proposition that the POI is a donor to the sample. Provided that the proposition is accepted by the court, activity related propositions can be put forward [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ]. It is common ground that the suspect and victim are living together:
      Hp: the suspect had vaginal intercourse with the victim.
      Hd: the suspect and the victim only had social contact.

      2.4.2 Bayesian network

      The Bayesian network is a representation of the definitions described above; Fig. 2, follows the scheme of Taylor et al. [
      • Taylor D.
      • Biedermann A.
      • Hicks T.
      • Champod C.
      A template for constructing Bayesian networks in forensic biology cases when considering activity level propositions.
      Forensic Sci. Int. Genet. 2018; 33: 136-146
      ], described in the supplement of Gill et al. [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ]. The top node in black represents the propositions and reflects the positions – as understood – of the prosecution and the defence respectively. Below, the blue nodes represent the sub-activities that take into account the various mechanisms by which DNA or vaginal mucosa could be transferred from the victim to the suspect. The probabilities that populate these nodes are either zero or one. If it is not disputed that the victim and the suspect cohabited, then we assign Pr = 1 for this state under both propositions. Below the sub-activity nodes sit the yellow nodes which contain transfer and accumulation probabilities; for the DNA nodes, the probability of presence/absence is defined by the average RFU value (log10RFU¯POI>x) where x is a variable threshold; i.e., if the value is above the threshold then it is present, whereas if it is below the threshold it is absent (Section 3.4.1); the presence/ absence of vaginal mucosa are conditioned upon the average RFU value (log10RFU¯POI) of the DNA (Section 3.4.3). Grey represents background nodes i.e. presence/absence of DNA or vaginal mucosa from unknown individuals. Red represents the results nodes, listing all possible outcomes that could be obtained from DNA/RNA profiling from the case. Likelihood ratios are calculated from the outcomes represented in the results nodes under the two alternative propositions. The BN was prepared using Hugin software (www.hugin.com). It was separately encoded using R v. 4.1.0 with formulae derived in Section 2.4.4. The programme (package DNARNA) can be downloaded from https://github.com/peterdgill/DNAVagMucosa along with the data file and user manual. The data output is available in Supplementary material 3, 4 and 5.
      Fig. 2
      Fig. 2Bayesian network of the case example introduced in .

      2.4.2.1 BN summary

      • 1)
        Either victim (V) and suspect (S) had sexual intercourse, or they only cohabited (blue nodes).
      • 2)
        If there is a positive RNA test for vaginal mucosa recovered from S penis then this occurred either from co-habitation (e.g., where common surfaces were touched) or from vaginal penetration (yellow nodes).
      • 3)
        If DNA from V was recovered from S penis, then this occurred either from co-habitation or from vaginal penetration (yellow nodes).
      • 4)
        In the second layer of yellow nodes: either vaginal mucosa RNA is present on S penis or it is absent. Similarly, DNA from the victim is either present or absent.
      • 5)
        Background vaginal mucosa RNA and background DNA (grey nodes) refer to material from unknown individuals.
      • 6)
        The results nodes, coloured red, calculate likelihood ratios from vaginal mucosa test results, DNA results and combined DNA/ vaginal mucosa test results.

      2.4.3 List of variables for Bayesian network analysis

      The notation used by Gill et al. [
      • Gill P.
      • Bleka Ø.
      • Roseth A.
      • Fonneløp A.E.
      An LR framework incorporating sensitivity analysis to model multiple direct and secondary transfer events on skin surface.
      Forensic Sci. Int. Genet. 2021; 53 (102509-102509)
      ] is followed, summarised in Table 1:
      Table 1Notations applied in the Bayesian network analysis.
      NotationDefinition
      tthe probability of direct transfer, persistence and recovery of DNA/RNA from the POI (under Hp only)
      t'the probability of direct transfer, persistence and recovery of DNA/RNA from an unknown contributor (under Hd only)
      bthe probability of background (DNA or RNA), based on observations, applied under both Hp and Hd. Background DNA is present from unknown sources and unknown activities. It can be described as ‘foreign’ (non-self). For further details we refer to Section 3.2 in
      • Gill P.
      • Bleka Ø.
      • Roseth A.
      • Fonneløp A.E.
      An LR framework incorporating sensitivity analysis to model multiple direct and secondary transfer events on skin surface.
      Forensic Sci. Int. Genet. 2021; 53 (102509-102509)
      .
      sthe probability of indirect transfer, persistence and recovery of DNA/RNA
      xdecision threshold value based upon log10RFU̅POI
      node VVaginal mucosa RNA test results (Bayesian network)
      node DDNA results (Bayesian network)
      node DVCombined DNA/ Vaginal mucosa RNA test results (Bayesian network)
      V+ or V-Positive (+) or negative (-) test for vaginal mucosa (Bayesian network)
      D+ or D-Presence (+) or absence (-) of DNA (Bayesian network)
      Suffixes are applied to identify the BN results node to which probabilities apply: e.g., tD>x refers to the probability of direct transfer, persistence and recovery of DNA of the POI (DNAPOI), and sD>x refers to the probability of indirect transfer, persistence and recovery of DNA of the POI where x is a threshold value of log10RFU̅POI. The terms are respectively abbreviated to tD and sD in subsequent text.
      The vaginal mucosa test result is always conditioned upon the quantity of DNA present, log10RFU̅POI, e.g., tV|D refers to the probability of direct transfer, persistence and recovery of vaginal mucosa of the POI conditioned on D; sV|D refers to the probability of indirect transfer, persistence and recovery of vaginal mucosa of the POI conditioned on D; The terms are respectively abbreviated to sV and tV in subsequent text.

      2.4.4 Formulae used to calculate likelihood ratios

      2.4.4.1 DNA results

      Recall the propositions.
      Hp: the suspect had vaginal intercourse with the victim.
      Hd: the suspect and the victim only had social contact.
      POI DNA is present, with or without an unknown contributor(s).
      Note that we do not consider an unknown individual in this network, because there is no dispute about who may or may not have committed the offence, and this is often the case in sexual assault cases [
      • Sijen T.
      • Harbison S.
      On the identification of body fluids and tissues: a crucial link in the investigation and solution of crime.
      Genes. 2021; 12: 1728
      ]. The background term is the same in the numerator and denominator, hence it cancels.
      Hp: Either there is indirect transfer and no direct transfer, or there is direct transfer with no indirect transfer or there is both indirect and direct transfer. The probability is PrE|Hp=sD1−tD+tD1−sD+tDsD1−bD
      PrE|Hp=sD1−tD+tD1−bD
      (2)


      Hd: There is indirect transfer only, with no background. The probability is PrE|Hd=sD1−bD and the corresponding LR is:
      LRD+=sD1−tD+tDsD
      (3)


      Because background cancels, the same formula is used if unknown contributors are present.
      POI is not present, with or without unknown contributor(s).
      Hp: No indirect transfer, and no direct transfer. The probability is PrE|Hp=1−sD1−tD.
      Hd: No indirect transfer: The probability is PrE|Hd= (1−sD)
      LRD−=1−tD
      (4)


      2.4.4.2 Vaginal mucosa test results

      Vaginal mucosa test positive
      Hp: Either there is indirect transfer and no direct transfer, or there is direct transfer with no indirect transfer or there is both indirect and direct transfer: The probability is PrE|Hp=sV1−tV+tV1−sV+tVsV
      PrE|Hp=(sV(1−tV)+tV)
      (5)


      Hd: Either there is indirect transfer (with or without background) or there is no indirect transfer with background present. The probability is PrE|Hd=sV1−bV+sVbV+bV1−sV
      PrE|Hd=sV+bV1−sV
      (6)


      LRV+=(sV(1−tV)+tV)sV+bV1−sV
      (7)


      Note that background (bV) does not cancel in this equation. Background refers to a positive test for vaginal mucosa from unknown sources/contributors. This is distinct from indirect transfer from prevalent (known) contributors, but in practice it is not possible to be able to distinguish if an observation V+ is from unknown or known individual(s) or from or a combination as the RNA test is not connected to the DNA profile. Furthermore, it is possible that vaginal mucosa could be detected in the absence of a DNA profile from the donor (defined in Section 2.4.4.3).
      Vaginal mucosa test negative
      Hp: There has been no transfer (with or without indirect transfer) and no background transfer
      PrE|Hp=1−sV1−tV1−bV
      (8)


      Hd: There has been no indirect transfer and no background transfer
      PrE|Hd=1−sV1−bV
      (9)


      LRV−=1−tV
      (10)


      2.4.4.3 Combining DNA results with vaginal mucosa test result (node DV)

      There is co-dependency between D and V (Section 3.4.3), hence V is conditioned upon D to calculate the joint probability:
      PrD>x,V=PrD>x×PrV|D>x
      (11)


      where x is a threshold value of log10RFU̅POI. There are four possible combined outcomes of log10LRDV (conditioning omitted for brevity) listed below, obtained by summation of (log10) results, where the term log10LRV is dependent upon D:
      log10LRDV=log10LRD+log10LRV
      (12)


      POI and vaginal mucosa: log10LRD+V+.
      POI only (no vaginal mucosa): log10LRD+V−.
      Vaginal mucosa only (no POI): log10LRD−V+.
      No POI or vaginal mucosa: log10LRD−V−.

      2.4.5 Distribution fitting

      Two kinds of distribution fittings were carried out: (a) with logistic regression the probability of success is modelled, where success is defined as the probability that the average RFU̅will reach a certain threshold (x) after a defined period of time. The longer the period of time between the deposition of the sample and the test, the lower the amount of DNA is recovered. (b) log-normal distributions were fitted to data where logistic models provided a poor fit. Here, the probability that the RFU̅ reaches a threshold (x) is modelled relative to the value of x. There is no time dependency; higher RFU̅ values were less commonly observed. To summarise:
      Direct transfer of DNA (tD) was modelled from Bayesian logistic regression for penile and fingernail swabs, using the R stan_glm function [
      • Muth C.
      • Oravecz Z.
      • Gabry J.
      User-friendly Bayesian regression modeling: a tutorial with rstanarm and shinystan.
      Tutor. Quant. Methods Psychol. 2018; 14: 99-119
      ] from package rstanarm, utilising MCMC to generate 4000 pairs of coefficients per logistic regression to find Prlog10RFU̅POI>x|Time where x is the threshold value tested and Time is the time-interval in hours between the transfer and collection of samples.
      Positive vaginal mucosa test from direct transfer (tV) was modelled with Bayesian logistic regression, using the stan glm function, with both Time and RFU̅POI as explanatory variables. Based upon the results (Section 3.4.3) only the latter was kept in the model to find Prlog10RFU̅POI<x. Data described in Section 3.4.3 were concurrently analysed using the glm function in R (results were consistent).
      Indirect transfer of DNA (sD) was modelled from log normal distributions of penile and fingernail swabs; also modelled for both direct (tD) and indirect (sD) transfer for boxershorts, where there is no time dependency. Log-normal distributions were fitted to the data using the R package fitdistrplus using function lnorm [
      • Gill P.
      • Bleka Ø.
      • Fonneløp A.E.
      RFU derived LRs for activity level assignments using Bayesian Networks.
      Forensic Sci. Int. Genet. 2022; 56102608
      ]. The method is described in detail in Supplementary material 2.
      To carry out sensitivity analysis, 4000× bootstraps (with replacement) were taken of datasets. For each bootstrap, a new set of log normal parameters (mean log and SD log) were calculated using the fitdistrplus R package using the lnorm function.
      The probability distributions were used to substantiate the Bayesian networks described in 2.4, 3.4. The programming of these networks was carried out with R-code using the formulae described in Section 2.4. The 4000 bootstrapped samples from log normal distributions were combined with 4000 MCMC estimates from Bayesian logistic regression and were used in sensitivity analysis, described by quantiles (0.025 – 0.975) to define the plausible range of LRs, as described by Gill et al. [
      • Gill P.
      • Bleka Ø.
      • Fonneløp A.E.
      RFU derived LRs for activity level assignments using Bayesian Networks.
      Forensic Sci. Int. Genet. 2022; 56102608
      ].

      3. Results

      3.1 DNA profiling

      A positive finding of very strong supporting evidence [

      ENFSI, E.N.F.S.I. guideline for evaluative reporting in forensic science, 2016.

      ] corresponds to DNA results that supports the proposition that the person of interest (POI) is a donor with sub-source LR ≥ 10 000, as described in Section 2.3.4. The DNA STR results and the number of positive samples, are presented in Table 2. Overall, 14 of the samples produced a single source DNA profile from the donor, five samples gave a single source profile matching the reference sample of the POI and the remaining 139 samples were DNA mixtures (two or three person mixtures), see Supplementary material 1, Table S2. Positive results were detected in 84 % (133) of all collected samples; 94 % (115) of the intimate contact and 50 % (18) of the non-intimate contact samples. The five single source profiles of the POI were detected in three fingernail swabs and two penile swabs collected immediately after intimate contact (0 h). In the remaining 128 positive samples, the POI was detected in mixtures together with DNA from donor or donor and an unknown contributor. The POI was the major contributor (> 60 % of the DNA profile) in 46 % (56) of the intimate contact samples, but never observed as the major contributor in a non-intimate contact sample.
      Table 2The percentage of positive samples (number of observations in brackets), POI is the major contributor (> 60% of the DNA profile) and unknown contributors are detected.
      Sample typeSampling timeAverage no. of hand washPositive samples (LR ≥ 10,000)POI major contributor (> 60 %)Unknown contributor detectedTotal number of samples
      Fingernail swabs00100 % (12)92 % (11)012
      12491 % (10)45 % (5)18 % (2)11
      187100 % (10)20 % (2)010
      24991 % (10)09 % (1)11
      361290 % (9)10 % (1)20 % (2)10
      Non-intimate contact50 % (6)033 % (4)12
      Penile swabs0100 % (12)92 % (11)8 % (1)12
      12100 % (12)50 % (6)8 % (1)12
      1891 % (10)36 % (4)011
      2491 % (10)36 % (4)18 % (2)11
      3682 % (9)27 % (3)9 % (1)11
      Non-intimate contact33 % (4)050 % (6)12
      Boxershorts0100 % (11)82 % (9)011
      Non-intimate contact67 % (8)025 % (3)12
      Overall, unknown contributors were detected in 11 % (18) of the samples (Table 3). Two boxershorts samples collected after non-intimate contact had the highest content where 10 % and 13 % of the profile could be explained by the unknown contributors. In one of the latter samples the unknown contributor was interpreted as DNA from a child of the donor and POI based on the number of alleles, RFU values and genotypes of present alleles; an LR calculation supported the proposition of DNA from donor and untyped child (of POI) rather than the alternative of DNA from donor and POI.
      Hp1: DNA from donor and POI, Hd1: DNA from donor and an unrelated unknown, Hd2: DNA from donor and untyped child (of POI).
      LR alt. 1: Pr (Evidence|Hp1) / Pr (Evidence|Hd1) = 1,545,732.
      LR alt. 2: Pr (Evidence|Hp1) / Pr (Evidence|Hd2) = 0.0005.
      In the remaining 16 samples unknown contributors represent a very small percentage of the DNA profiles, i.e., 3.6 % or lower.
      Table 3Samples with unknown contributors divided into sample type and category, number of unknown alleles (different from donor and POI) and the percentage of unknown contributor in the DNA profiles (N = 18).
      Sample typeSample categoryTotal number of samples with unknown contributorNumber of unknown alleles (lowest – highest observation)% of unknown contributor in the DNA profile (lowest – highest observation)Average % of unknown contributor to DNA profile
      Fingernail swabsIntimate contact41–127.5E-07 – 3.61.8
      Non-intimate contact42–110.064 – 3.11.66
      Penile swabsIntimate contact41–21.9E-09 – 0.550.26
      Non-intimate contact31–60.052 – 1.91.18
      BoxershortsIntimate contact0000
      Non-intimate contact31–180.13 – 137.68
      The log10RFU̅POI decreases with time of sampling (Fig. 3). Statistical analysis shows that the log10RFU̅POI was significantly higher when intimate contact samples were compared with non-intimate contact samples, and likewise with penile swabs compared to fingernail swabs in intimate contact samples (t-tests, p < 0.00004).
      Fig. 3
      Fig. 3Box plot displaying the distribution of log10RFU̅POI in positive samples divided into time of samples collection post intimate contact and non-intimate contact samples. The intimate contact boxershorts samples are included in the “0 h” category, N = 133. (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).
      The distribution of sub-source LR values (the probability of the DNA result given that the DNA is from POI rather than from an unknown individual who is unrelated to POI, further described in Section 2.3.4) is shown for all samples in Fig. 4. The LR values for intimate contact samples are higher than non-intimate contact samples, but the value does not differ much between the different time points of the intimate contact samples. There is a higher variation in the LR values for the later time points, especially for the fingernail swabs.
      Fig. 4
      Fig. 4Box plot displaying the sub-source LR values (the probability of the DNA result given that the DNA is from POI rather than from an unknown individual, unrelated to POI) for all samples except the single source profile samples of donor. The Samples are classified by time of sampling (post intimate contact) and non-intimate contact samples. The intimate contact boxershorts samples are included in the “0 h” category, N = 144. (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).

      3.2 RNA profiling

      The mRNA results, based on the scoring categories described in Section 2.3.3, are presented in Table 4. Overall, a positive test for vaginal mucosa was observed in 43% (68) of all collected samples; 55 % (67) of the intimate contact samples and 3 % (1) of the non-intimate contact samples. The success of the detection of vaginal mucosa decreases with time from 100 % at 0 h to 20 % and 27 % at 36 h for fingernail and penile swabs respectively. The success rate is generally higher in penile swabs than fingernail swabs. In intimate contact samples the occurrence of sporadic detections is similar for the different time points. In non-intimate contact samples, vaginal mucosa was not detected except for one sample from a pair of boxershorts. In this particular sample DNA from the person of interest (POI) was also detected (see Section 3.3). A total of 56 samples scored sporadic detection for vaginal mucosa. Of these, 36 samples (20 intimate contact samples and 16 non-intimate contact samples) were due to the detection of the MUC4 marker (Supplementary material 1, Fig. S1).
      Table 4The results of mRNA vaginal mucosa markers (number of observations in brackets) divided into different sample types, sampling time, average number of hand wash and the detection of other body fluids.
      Sample typeSampling timeAverage no. of hand washmRNA Vaginal mucosa detected (≥ 50 % rule)mRNA Vaginal mucosa sporadically detectedmRNA Vaginal mucosa NOT detectedDetection of other body fluids (≥ 50 % rule)Total number of samples
      Fingernail swabs00100 % (12)0025 % (3)12
      12445 % (5)45 % (5)9 % (1)011
      18720 % (2)50 % (5)30 % (3)010
      2499 % (1)55 % (6)36 % (4)9 % (1)11
      361220 % (2)50 % (5)30 % (3)010
      Non-intimate contact042 % (5)58 % (7)8 % (1)12
      Penile swabs0100 % (12)0067 % (8)12
      1292 % (11)08 % (1)42 % (5)12
      1863 % (7)27 % (3)9 % (1)36 % (4)11
      2427 % (3)73 % (8)036 % (4)11
      3627 % (3)45 % (5)27 % (3)27 % (3)11
      Non-intimate contact025 % (3)75 % (9)17 % (2)12
      Boxershorts082 % (9)18 % (2)064 % (7)11
      Non-intimate contact8 % (1)75 % (9)17 % (2)17 % (2)12
      The distribution of the RFU values of each mRNA vaginal mucosa marker is presented in Fig. 5. Except for samples collected immediately after intimate contact (0 h), which produced higher RFUs, the RFUs are similar within each marker across the different time points of sample collection and for non-intimate contact samples. The same trend was observed when different kinds of samples were considered, however, the MYOZ1 marker shows higher RFU values in penile swabs than fingernail swabs (Supplementary material 1, Fig. S2).
      Fig. 5
      Fig. 5Box plot displaying the (log10) RFU values of the mRNA vaginal mucosa markers, including all parallels (inputs of cDNA), in intimate contact samples collected at different time points and in non-intimate contact samples. The intimate contact boxershorts samples are included in the “0 h” category. (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).
      The detection of the different body fluids/ cell types is presented in Table 5. After vaginal mucosa, semen was the most frequent body fluid detected with a detection rate of 21 % (see Supplementary material 1, Table S3). While vaginal mucosa markers were detected in 43 % of all samples analysed, the female marker was only reported in 13 % of the samples where the majority of samples were collected at 0 h or in boxershorts after intimate contact. The male marker had a higher success rate and was detected in 48 % of all samples. In a fingernail swab collected after non-intimate contact from one of the participants the saliva marker STATH was detected in all three parallel analyses. In addition MUC4 and BPIFA1 (nasal mucosa marker) were detected at 10X RFU. STATH was therefore interpreted as co-expressed with nasal mucosa, i.e. “observed and fits with” the presence of nasal mucosa (and not saliva) as described in Lindenbergh et al. [
      • Lindenbergh A.
      • Maaskant P.
      • Sijen T.
      Implementation of RNA profiling in forensic casework.
      Forensic Sci. Int. Genet. 2013; 7: 159-166
      ]. Up to four body fluids were detected in two of the samples (fingernail and penile swab collected at 0 h after intimate contact from one participant) where vaginal mucosa, semen, blood and saliva were detected in addition to male and female markers.
      Table 5The percentage of samples (exact numbers in brackets) in which the various body fluids, male and female markers were “detected” according to scoring rules. The last column shows the total number of samples analysed in each category.
      Sample typeSampling timeVaginal mucosaBloodSemenSalivaMenstrual secretionNasal mucosaFemale markerMale markerTotal number of samples
      Fingernail swabs0100 % (12)8 % (1)25 % (3)8 % (1)0033 % (4)33 % (4)12
      1245 % (5)00000036 % (4)11
      1820 % (2)00000020 % (2)10
      249 % (1)9 % (1)09 % (1)00018 % (2)11
      3620 % (2)00000040 % (4)10
      Non-intimate contact000008 % (1)033 % (4)12
      Penile swabs0100 % (12)25 % (3)33 % (4)25 % (3)0067 % (8)75 % (9)12
      1292 % (11)042 % (5)0008 % (1)67 % (8)12
      1863 % (7)036 % (4)0009 % (1)55 % (6)11
      2427 % (3)036 % (4)0009 % (1)64 % (7)11
      3627 % (3)027 % (3)000045 % (5)11
      Non-intimate contact008 % (1)008 % (1)033 % (4)12
      Boxer-shorts082 % (9)18 % (2)64 % (7)00055 % (6)91 % (10)11
      Non-intimate contact8 % (1)017 % (2)000058 % (7)12

      3.3 Association of RNA and DNA results

      Vaginal mucosa was only detected in samples where POI was detected. Vaginal mucosa and DNA from POI was detected in one non-intimate contact sample collected from a pair of boxershorts. However, vaginal mucosa was not detected in any of the other non-intimate contact samples with or without DNA from POI (Supplementary material 1, Figs. S3 and S4).
      The log10RFU̅POI combined with detection of vaginal mucosa is presented in Fig. 6. It is higher in the intimate contact samples compared to non-intimate contact samples in penile swabs and boxershorts. The log10RFU̅POI was significantly higher when comparing samples with vaginal mucosa detected compared to the non-detected samples (none or sporadic detection of mRNA markers), according to the t-test (p < 0.0004). The RFU̅POI in non-intimate contact samples are at the same level or lower than the 18–36 h fingernail samples where vaginal mucosa was not detected.
      Fig. 6
      Fig. 6Scatter plots of log10RFU̅POI in positive samples (LR ≥ 10 000) where vaginal mucosa was detected (blue) or not detected (light brown), i.e. none or sporadic detection of mRNA vaginal mucosa markers. Samples are divided into boxershorts, fingernail swabs and penile swabs collected at different time points post intimate contact and non-intimate contact samples. The intimate contact boxershorts samples are included in the “0 h” category, N = 133. (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).

      3.4 Characterisation of variables used in the Bayesian network

      3.4.1 Direct transfer of POI DNA (tD)

      Penile and fingernail swabs
      Logistic regressions were calculated as described in Section 2.4.5, comparing time since intercourse versus Prlog10RFU̅>x recoveries for penile and fingernail swabs, where x is a threshold value (Fig. 7).
      Fig. 7
      Fig. 7Logistic regressions of a) penile swabs (left), b) fingernails swabs (right). Time since intercourse vs Prlog10RFU¯POI>x, showing probability of DNA transfer, persistence and recovery for a range of threshold values x. (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).
      The plot shows that substantial amounts of POI DNA are recovered from penile swabs, even after 36 h where the probability is 0.45 of recovering log10RFU¯POI>4. Fingernails do not retain as much DNA; the equivalent recovery Pr = 0.45 for log10RFU̅>4 is achieved at 15 h while this probability is approximately 0 at 36 h.
      Boxershorts
      There is also a high expectation of direct transfer with boxershorts. There is no time dependency in this model; a log normal distribution was calculated as described in Section 2.4.5. Prlog10RFU¯POI>4≈1, with a subsequent fall in the probability for increasing log10RFU̅POI≈7 to Pr = 0.006 (Fig. 8).
      Fig. 8
      Fig. 8Log normal distribution fit for boxershorts for direct transfer, showing RFU̅POI vs.tD.

      3.4.2 Indirect transfer of DNA from the POI (sD)

      Log normal distributions (Section 2.4.5) were fitted using the fitdistrplus R package (Fig. 9) for penile and fingernail swabs, and boxershorts (Supplementary material 2). Indirect transfer rates are much lower compared to direct transfer, noting that sD < 0.1 for RFU̅>2000, which has a corresponding effect upon the likelihood ratios observed (described in Section 3.4.5). Penile swabs show the lowest levels of indirect transfer, and boxershorts the highest, with fingernail swabs as intermediate.
      Fig. 9
      Fig. 9Log normal distribution fits for penile and fingernail swabs and boxershorts for indirect transfer of POI DNA, showing RFU̅POI vs. sD (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).

      3.4.3 Detection of vaginal mucosa RNA (tv) after direct contact

      There are two potential explanatory variables, time since intercourse (Time) and the amount of DNA (RFU̅POI), that may impact probability of detection of vaginal mucosa (tv) as the dependent variable.
      Logistic regressions were carried out for both penile and fingernail swabs (Table 6). In summary, the data showed no statistically significant impact of Time upon the dependent variable (p > 0.16). However, there was a strong effect with log10RFU̅POI (p ≤ 0.01).
      Table 6Summary from logistic regressions of penile and fingernail swabs where positive/ negative vaginal mucosa test is the dependent variable and explanatory variables are Time and log10RFU̅POI. Statistically significant results are italised.
      Penile swabsEstimateStd. Errorz valuePr (>|z|)
      (Intercept)-8.674.06-2.1350.03
      Time-0.070.05-1.420.16
      RFU2.350.812.9090.004
      Fingernail swabsEstimateStd. Errorz valuePr (>|z|)
      (Intercept)-20.117.05-2.850.004
      Time0.050.060.750.46
      RFU4.731.563.020.003
      Comparison of detection of vaginal mucosa on penile swabs, underneath fingernails and boxershorts after direct contact.
      Based upon the impact study in the previous section, Bayesian logistic regression of tVwas carried out with log10RFU̅POI as the only explanatory variable in the three experiments (penile, fingernail swabs, and boxershorts). The observed distributions are similar (Fig. 10); 0.025 and 0.975 quantiles for penile swabs overlap medians for fingernail swabs and boxershorts. Quantiles were calculated directly from the 4000 bootstraps used to calculate the logistic regression coefficients.
      Fig. 10
      Fig. 10Logistic regression showing probability of vaginal mucosa transfer (tv) conditioned upon log10RFU¯POI : comparison of penile (with quantiles) and fingernail swabs, and boxershorts. (For interpretation of the references to colour in this figure, the reader is referred to the online version of this article.).
      If Hp is true, at low log10RFU̅POI values, e.g. log10RFU¯POI<3, vaginal mucosa is rarely detected (tv < 0.1). Conversely, at log10RFU¯POI>6, vaginal mucosa will usually be detected (tv > 0.9), with intermediate probabilities of (tv) between the two extremes.

      3.4.4 Detection of vaginal mucosa RNA after indirect contact (sv) and background levels (bv)

      Indirect transfer (from a known individual) and background (from an unknown individual) cannot be distinguished for an observation of vaginal mucosa in evidential material. By combining the results in our study (non-intimate contact penile swabs) with another study [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ], consisting of total 23 penile swabs, there was only a single observation (1/23) of a positive test for vaginal mucosa. We assigned this value to both sv and bv and assumed uniform distributions relative to RFU̅POI. The same value was also assigned to fingernail swabs and boxershorts, due to lack of published data in the literature.

      3.4.5 Activity likelihood ratio (LRDV) analysis: Penile swabs

      In Section 2.4.4, the formulae for the BN (Fig. 2) are described. There are two sets of results that are generated: Vaginal Mucosa RNA test results (node V) and the DNA results (node D). Likelihood ratios can be generated for each, conditioned under the respective Hp/Hd propositions. The two sets of results can also be calculated as a joint probability (node DV) as described in Section 2.4.4.3, Eq. 12.
      It is of interest to determine the effect on the likelihood ratio under various experimental conditions: a DNA test is always carried out as standard; but the question is what is the overall impact on the likelihood ratio result when vaginal mucosa is tested either alone or in conjunction with the DNA test? The experimental tests carried out, and their possible outcomes are listed as follows:
      • a)
        DNA test and vaginal mucosa test is carried out where:
        • a.
          Vaginal mucosa test is positive (D+V+)
        • b.
          Vaginal mucosa test is negative (D+V-)
      • b)
        The DNA test is carried out alone; no vaginal mucosa test is carried out (D+)
      • c)
        Contribution of the vaginal mucosa test alone and:
        • a.
          Vaginal mucosa test is positive (V+)
        • b.
          Vaginal mucosa test is negative (V-)
      log10LRs for BN results nodes D, V and DV, for penile swabs, were calculated (Table 7) using the formulae listed in Section 2.4.4. For node D, tDand sD are conditioned on values of x, where x is the RFU threshold and Times = 0 h and 35 h. Node V is not time-dependent and is conditioned only upon values of log10RFU¯POI. The purpose was to compare the outcomes where nodes D and V were evaluated separately and combined. Sensitivity analysis was carried out using bootstrapping and Bayesian logistic regression (Section 2.4.5) - quantiles ranging from 0.025 to 0.975 were calculated. Tables for fingernail swabs and boxershorts are available in Supplementary material 4 and 5 respectively.
      Table 7log10LR analysis for penile swabs taken either Time = 0 h or 35 h after intercourse relative to x threshold. LR equations are listed in Section 2.4.4. Sensitivity analysis was carried out as described in Section 2.4.5 to generate quantiles. tD values are medians from bootstrapped data.
      AD+V+log10LRD+V+ Quantiles
      Time (h)xtD| (Time,RFU>x)0.0250.050.250.50.750.950.975
      020.99-0.16-0.139-0.10.00.10.30.3
      030.990.60.60.91.21.63.85.7
      040.98333451012
      050.916679111515
      060.65991113151717
      3520.85-0.2-0.2-0.10.00.10.20.3
      3530.850.50.60.91.11.63.85.6
      3540.4522345912
      3550.05456791314
      3560.00156810121414
      BD+V-log10LRD+V- Quantiles
      Time (h)xtD|(Time,RFU>x)0.0250.050.250.50.750.950.975
      020.990.100.130.190.250.320.450.52
      030.990.70.71.01.21.63.85.6
      040.9822345912
      050.91445791314
      060.6566810121414
      3520.850.10.10.20.20.30.40.5
      3530.850.60.70.91.11.63.85.5
      3540.4522234811
      3550.05234681212
      3560.001235791111
      CD+onlylog10LRD+ Quantiles
      Time (h)xtD|(Time,RFU>x)0.0250.050.250.50.750.950.975
      020.990.110.130.190.250.320.450.52
      030.990.70,.81.01.21.63.85.6
      040.9822345912
      050.915568101414
      060.65881012141516
      3520.850.10.10.20.20.30.40.5
      3530.850.60.7091.21.63.85.6
      3540.4522335912
      3550.05345681213
      3560.0014579111313
      Dnode V onlylog10LRV Quantiles
      log10RFUtV|(RFU)0.0250.050.250.50.750.950.975
      Vaginal mucosa recovered (V+)20.002-0.3-0.3-0.3-0.3-0.2-0.10.0
      30.03-0.3-0.2-0.2-0.10.10.30.4
      40.280.20.30.50.60.70.80.8
      50.840.90.91.01.01.01.01.0
      60.991.01.01.11.11.11.11.1
      Vaginal mucosa not recovered (V-)20.0020.00.00.00.00.00.00.0
      30.03-0.1-0.10.00.00.00.00.0
      40.28-0.3-0.3-0.2-0.1-0.1-0.1-0.1
      50.84-1.2-1.1-0.9-0.8-0.7-0.5-0.5
      60.99-2.8-2.7-2.1-1.8-1.5-1.2-1.1
      A: log10LRD+V+ for BN combined node DV, Eq. 12: DNA and vaginal mucosa recovered (D+V+), (3), (7).
      B: log10LRD+V− for BN combined node DV, Eq. 12: DNA recovered; no vaginal mucosa recovered (D+V-), (3), (10).
      C: log10LRD+ for BN node D: DNA recovered; vaginal mucosa not tested (D+ only), Eq. 3.
      D: log10LRV for BN node V: Vaginal mucosa tested; DNA not tested (V+ and V- only), (7), (10).
      Table 7C shows likelihood ratios for node D when only DNA results are available. Table 7D shows likelihood ratios for node V when only vaginal mucosa results are available. Table 7 A and 7B shows likelihood ratios for node DV when both DNA and vaginal mucosa are tested.

      3.4.5.1 Node V

      Results of log10LRV+ (Table 7D) show that when there is a positive test for vaginal mucosa, at low log10RFU̅POI=2−3, the evidence weakly favours the Hd proposition (this is because background and indirect transfer are more probable), whereas at higher log10RFU̅POI=4−6 values, the evidence weakly favours Hp,: maximum log10LRV+=1.1. Conversely, if the test is negative, for log10LRV− the evidence is neutral at low log10RFU̅POI=2−3, but a negative test of vaginal mucosa at higher log10RFU̅POI=4−6 reduces the log10LRV+, supporting Hd.

      3.4.6 Impact of vaginal mucosa RNA test BN node (V) on the evaluation of the combined DNA/RNA results

      In combination, values of log10LRDV are overwhelmingly influenced by node D. The presence/absence of vaginal mucosa has very low impact on the combined log10LRDV results: either log10LRD+V+ is increased or log10LRD+V− is decreased by an approximate order of magnitude relative to log10LRD for log10RFU¯POI>4 (Table 7A-C). Respective quartiles of the sensitivity analysis overlap between the different D+V+ , D+V- categories.
      A precise understanding of the time since intercourse is not essential to provide a likelihood ratio. However, the longer the time interval between activity and collection, the lower the likelihood that high quantities of DNA will be recovered (Fig. 7, Table 7). For example, if Hp = true, at Time = 0 h (D+V+), tD|(log10RFU¯>5)=0.091, reducing to tD|(log10RFU¯>5)=0.05 at Time = 35 h, i.e. in casework, the natural loss of DNA over time reduces log10RFU̅POI, which in turn reduces the log10LR that can be realistically achieved.
      If vaginal mucosa is recovered, and no DNAPOI is recovered, then the log10LRD−V+ is always less than one (data not shown).

      3.4.7 Comparison of activity LRs from penile and fingernail swabs, and boxershorts

      A comparison of median log10LRD for three different evidence types is shown in Table 8. Recovery of DNA from boxershorts is not time dependent. For log10RFUPOI > 4, the likelihood ratios of boxershorts are higher than for penile swabs, and lower still for fingernail swabs; the success rates of retrieving higher log10RFUPOI from fingernails is much reduced for Time = 35 h, compared to penile swabs. There are practical limitations upon achieving high LRs for Time = 35 h. Consequently, boxershorts are a useful source of evidence, especially if the suspect has bathed before samples can be collected. A comprehensive view of the data, including sensitivity analysis is provided in Supplementary material 3, 4 and 5.
      Table 8log10LRD+ (median) analysis for BN node D: "DNA Results" (Fig. 2), Eq. 3, comparing results from penile and fingernail swabs taken either Time = 0 h or 35 h after intercourse. Results for boxershorts are not time dependent but can be compared with other results at Time = 0 h.
      Penile swabsFingernail swabsBoxershorts
      Time (h)xtD| (Time,RFU>x)log10LRD+ (median)tD| (Time,RFU>x)log10LRD+ (median)tD| (RFU>x)log10LRD+ (median)
      020.990.250.990.210.18
      030.991.20.980.910.7
      040.9840.97314
      050.9180.860.8610
      060.65120.2180.213
      3520.850.20.900.2
      3530.851.20.580.7
      3540.4530.011
      3550.0560.00032
      3560.00190.0016

      3.5 Case example

      To show how the data obtained in this study can be used to address activity related propositions in casework, a case example is introduced using Bayesian network to weight the evidence.
      A 30 year old woman (victim) had been living in a flat share for six months with a 30 year old man (suspect). They had separate bedrooms, but shared the living room, kitchen and bathroom. One morning the victim called the police stating that the suspect raped her while she was sleeping, i.e. vaginal penetration without her consent. According to the victim the activity stopped when she woke up, and the suspect left her room. The police arrested the suspect later that day. The suspect denied any sexual contact. No other individual was suspected of the offence.

      3.5.1 The propositions

      Hp: the suspect had vaginal intercourse with the victim.
      Hd: the suspect and the victim only had social interaction via cohabitation.

      3.5.2 The forensic examination

      Forensic examination of the victim and the suspect was performed. Penile swabs, fingernail swabs and boxershorts were analysed. Two sets of hypothetical findings (A and B) are discussed:
      A: Samples collected approximately 15 h after alleged offence
      • 1.
        DNA from the victim was detected on the penile swab (shaft), fingernail swabs and the boxershorts collected from the suspect. The log10RFU̅POI=5,4,5 respectively for each item of evidence.
      • 2.
        Positive test for vaginal mucosa was detected by mRNA analysis on all items
      • 3.
        No spermatozoa or seminal fluid was detected in the intimate samples collected from the victim.
      • 4.
        No DNA of the suspect was detected in the samples from the victim (DNA STR analysis)
      B: Samples collected approximately 25 h after alleged offence
      • 1.
        DNA from the victim was detected on the penile swab (shaft), fingernail swabs and the boxershorts collected from the suspect. The log10RFU̅POI=4,3,5 respectively, for each item of evidence.
      • 2.
        Negative test for vaginal mucosa by mRNA analysis on all three items.
      • 3.
        No spermatozoa or seminal fluid was detected from the intimate samples collected from the victim.
      • 4.
        No DNA of the suspect was detected in the samples from the victim (DNA STR analysis)
      Provided the court accepts the prosecution proposition that the defendant is present as a donor of DNA on the items of evidence [
      • Gill P.
      • Hicks T.
      • Butler J.M.
      • Connolly E.
      • Gusmao L.
      • Kokshoorn B.
      • Morling N.
      • van Oorschot R.A.H.
      • Parson W.
      • Prinz M.
      • Schneider P.M.
      • Sijen T.
      • Taylor D.
      DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
      Forensic Sci. Int. Genet. 2020; 44102186
      ] then activity level propositions may be addressed using the Bayesian network described in Section 2.4.
      However, before any analysis is carried out, the case circumstances can be used to help investigators to assess their prior expectations of the possible outcome of an analysis if Hp = true or if Hd = true. The most important dependency is the time interval between the offence and the collection of samples. Look-up tables in Supplementary material 3, 4 and 5 provide a list of direct transfer probabilities: tD|(log10RFU̅POI,Time=0−35h,Hp) for penile swabs, fingernail swabs and boxershorts. These data are also plotted in Fig. 7, Fig. 8. To illustrate the case specific example outlined above, expected outcomes for Time = 15 h and 25 h are detailed in Table 9.
      Table 9Table of expected outcomes where Time = 15 h and 25 h between offence and collection of samples. tD|(log10RFU̅POI>x,Time,Hp) is the probability of direct transfer. Listed in columns are log10LRDV median values calculated for positive/negative vaginal mucosa test (D+V+ and D+V-). The grey shaded cells provide expectation of the outcome (> log10RFU̅POI) of 50 % of trials (penile swabs), 44 % fingernail swabs, and 86 % boxershorts (Time = 15 h). Boxershorts do not have time dependency.
      Likelihood ratios are similar between the three different evidence types conditioned on a given log10RFU̅POI value; the time interval of sample collection has little effect upon the log10RFU̅POI conditioned LR. However, the actual LR which can be achieved is an entirely different consideration. This is dependent upon tD. The shaded cells in Table 9 show the achievable LR relative to tD|log10RFU¯POI>x chosen to be approximately 0.5 or higher. For example, for fingernail swabs taken Time = 15 h, tD|(log10RFU¯POI>4)=0.44, so it is a reasonable expectation to observe log10LRD+V+≈3, but the log10LR is less likely to exceed 5 since tD|(log10RFU¯POI>5)=0.06.
      A summary of the LRs given the results analysis of the exemplar case are also highlighted by the shaded cells in Table 9. Comparing the two sets of hypothetical findings (A and B) in the given case example, log10LRD+V+ = 8, 3 and 11 for penile swab, fingernail swab and boxershorts respectively for set A (Times = 15 h, positive test for vaginal mucosa and log10RFU̅POI=5,4,5 for each item respectively). In the case of set B, log10LRD+V− = 3, 0.8 and 9 for penile swab, fingernail swab and boxershorts respectively (Time = 25 h, negative test for vaginal mucosa and log10RFU̅POI=4,3,5 respectively). Since recovery from boxershorts are not time dependent, they can provide a good source of evidence. Values of log10LRD+V− are shown concurrently; there is an order of magnitude reduction for penile and fingernail swabs; two orders for boxershorts.

      4. Discussion

      An understanding of transfer, persistence, prevalence and recovery (TPPR) of DNA and mRNA body fluid markers are important to decide the examination strategy based upon expectation of findings, given the alternative propositions. Samples may be prioritised, based on the time between the alleged assault, the medical examination and collection of samples. In this study, we present the results of DNA and mRNA body fluid analysis from samples collected from 12 couples post intimate contact and after non-intimate contact. It was challenging to recruit participants for this study, as the experiment comprised collection of “personal samples”. The dataset presented in this paper consists of 158 samples (DNA and mRNA profiles), including 122 intimate contact samples and 36 non-intimate contact samples.

      4.1 Recovery of DNA

      The positive findings in the description of the data in this study (3.1, 3.2, 3.3) correspond to the value of the evidence supporting the prosecution's proposition that the person of interest (POI) is the donor to the DNA profile with LR ≥ 10 000. Positive results were detected in 94 % of the intimate contact samples and 50 % of the non-intimate contact samples. As there was no restriction to social contact, the positive findings on non-intimate contact fingernail swabs could be a result of either direct or indirect transfer of cellular material. Positive results were detected in 33 % and 67 % of the non-intimate contact penile and boxershorts samples respectively, and attributed to indirect transfer. Full profiles of the POI, i.e., 46 alleles, were detected in some of the non-intimate contact penile and boxershorts samples. This highlights that the evaluation of DNA evidence must be carried out with care when family members or co-inhabitants are involved, as there are likely to be several sources of their DNA in the household that can be innocently transferred. This is why proper formulation of propositions using the hierarchy of propositions framework and analysis of results using the Bayesian networks are needed to provide the value of the evidence.
      The sub-source LR threshold of 10 000 was used to describe positive findings (3.1, 3.2, 3.3). This is in line with case-work practice where a lower reporting threshold is applied, e.g., 10 000 as concluded in a validation study [
      • Gill P.
      • Bleka O.
      • Hansson O.
      • Benschop C.
      • Haned H.
      Forensic Practioner’s Guide to the Interpretation of Complex DNA Profiles.
      Academic Press, 2020: 304-306
      ]. Ten DNA mixtures in the dataset had an LR less than 10 000 with a RFU̅POI ranging from 0 to 3617, Supplementary material 1, Fig. S5. Background DNA (from unknown contributor(s)) was detected in 13.0% of boxershorts samples, 12.1 % of fingernail swabs and 10.1 % of penile swabs. However, presence/ absence of background DNA does not impact the Bayesian network (BN) in this study because it cancels out in the LR calculation. If the identity of the perpetrator was in dispute under Hd, the probability of background would have an impact on the LR. The percentage of unknown alleles may be lower due to Covid restrictions as the experiments were carried out during the pandemic. However, the results of unknown contributors in fingernail samples are in line with the findings of Malsom et al. [
      • Malsom S.
      • Flanagan N.
      • McAlister C.
      • Dixon L.
      The prevalence of mixed DNA profiles in fingernail samples taken from couples who co-habit using autosomal and Y-STRs.
      Forensic Sci. Int. Genet. 2009; 3: 57-62
      ] who detected unknown alleles in 16.7% of samples in a study from 12 couples.

      4.2 Recovery of vaginal mucosa

      It was observed that the detection of vaginal mucosa decreased rapidly within the first day of sample collection post intimate contact, but was still detected up to 36 h in some samples. Blackman et al. [
      • Blackman S.
      • Stafford-Allen B.
      • Hanson E.K.
      • Panasiuk M.
      • Brooker A.-L.
      • Rendell P.
      • Ballantyne J.
      • Wells S.
      Developmental validation of the ParaDNA® body fluid ID system – a rapid multiplex mRNA-profiling system for the forensic identification of body fluids.
      Forensic Sci. Int. Genet. 2018; 37: 151-161
      ] also reported a detection of the mRNA vaginal mucosa marker CYP2B7P1 in a post-coital penile swab collected from glans 36 h after intimate contact. In the samples collected at 12 h after intimate contact or later, DNA from the POI was frequently observed (positive findings) although the mRNA test for vaginal mucosa provided a negative result. This is expected as RNA is in general less stable than DNA [
      • Sijen T.
      Molecular approaches for forensic cell type identification: On mRNA, miRNA, DNA methylation and microbial markers.
      Forensic Sci. Int. Genet. 2015; 18: 21-32
      ]. However the findings of this study demonstrate that the RFU̅POI after intimate contact is significantly higher compared to the RFU̅POI after social contact, also at the later time points. This forms the basis for our model, where the vaginal mucosa results have a lower impact on the resulting LR. The low impact of the vaginal mucosa test result was unexpected but can be explained by the relative high level of false negatives, the RFU̅POI difference in the intimate contact versus non-intimate contact samples and the relatively high probability assigned for a positive test in the non-intimate contact samples which was applied to both indirect transfer (sV) and background (bV). The probability of a positive test for vaginal mucosa from non-intimate contact samples was set to 1/23. This number is based on the observations in our dataset combined with data in another study [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ]. Ideally, a larger sample size is needed to improve the assignment of the probability. A strong association was found between a positive test for vaginal mucosa and RFU̅POI. In the logistic regression considering both time and RFU̅POI, the time of collection did not have a significant impact as an explanatory variable on the presence or absence of vaginal mucosa (Table 6). This can be explained by the strong association between a positive vaginal mucosa test and RFU̅POI, and that RFU̅POI decreases with time. The probability of direct transfer of DNA (tD) reduces with time, resulting in lower detection rates of vaginal mucosa, along with reduced likelihood ratios that can be realistically achieved (Fig. 7, Table 7).
      It is possible that other body fluids, e.g. saliva, will provide similar results of RFU̅POI to vaginal mucosa. This has not been investigated in the present study; it would not impact the type of case discussed here as transfer of saliva is not relevant. For different kinds of cases, future research could include other body fluids and expand the BN to incorporate e.g. saliva.
      There was no restriction in this study on hand washing (Table 2 and 4). Although hand washing may have an impact upon recovery from underneath fingernails, the nails appear to offer protection against loss of mRNA markers, as well as DNA.
      Vaginal mucosa was not detected in any of the non-intimate contact samples except for one boxershorts sample. In this sample DNA from the POI was also detected (RFU̅POI=387). Assuming that sampling was carried out according to instructions, i.e. no intimate contact for minimum the last 72 h (3 days), the positive finding could be a result of: (a) persistence on the penis post 72 h, (b) indirect transfer of vaginal mucosa cells from e.g., the hands or (c) false positive result. The threshold of 72 h since last intimate contact seems reasonable, since no positive findings of DNA were detected after 48 h in the retrospective study of Fonneløp et el. [
      • Fonneløp A.E.
      • Johannessen H.
      • Heen G.
      • Molland K.
      • Gill P.
      A retrospective study on the transfer, persistence and recovery of sperm and epithelial cells in samples collected in sexual assault casework, Forensic Science.
      Forensic Sci. Int. Genet. 2019; 43
      ] and no positive mRNA results were found in non-intimate contact samples in the conducted pilot study (Section 2.2.2). Van den Berge et al. [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ] also observed vaginal mucosa in one penile swab where the DNA analysis showed a single male DNA profile. The mRNA multiplex used in this study contained three vaginal mucosa markers: MUC4, MYOZ1 and CYP2B7P1. The MUC4 marker was the most sensitive marker, but also the least specific. MUC4 was detected in 44 % of the non-intimate contact samples, and most frequently in boxershorts samples (Supplementary material 1, Fig. S1). Our findings support the conclusion of van den Berge et al. [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ] that MUC4 should be considered with care when penile swabs are analysed. They found that MUC4 marker was detected in 46 % of penile swab samples where DNA analysis proved single source male profiles. MUC4 is a mucin secretion marker which is known to also be present at high abundance in other body fluids, especially saliva/buccal swabs and nasal mucosa [
      • Haas C.
      • Hanson E.
      • Anjos M.J.
      • Ballantyne K.N.
      • Banemann R.
      • Bhoelai B.
      • Borges E.
      • Carvalho M.
      • Courts C.
      • De Cock G.
      • Drobnic K.
      • Dotsch M.
      • Fleming R.
      • Franchi C.
      • Gomes I.
      • Hadzic G.
      • Harbison S.A.
      • Harteveld J.
      • Hjort B.
      • Hollard C.
      • Hoff-Olsen P.
      • Huls C.
      • Keyser C.
      • Maronas O.
      • McCallum N.
      • Moore D.
      • Morling N.
      • Niederstatter H.
      • Noel F.
      • Parson W.
      • Phillips C.
      • Popielarz C.
      • Roeder A.D.
      • Salvaderi L.
      • Sauer E.
      • Schneider P.M.
      • Shanthan G.
      • Court D.S.
      • Turanska M.
      • van Oorschot R.A.
      • Vennemann M.
      • Vidaki A.
      • Zatkalikova L.
      • Ballantyne J.
      RNA/DNA co-analysis from human menstrual blood and vaginal secretion stains: results of a fourth and fifth collaborative EDNAP exercise.
      Forensic Sci. Int. Genet. 2014; 8: 203-212
      ,
      • Jakubowska J.
      • Maciejewska A.
      • Pawlowski R.
      • Bielawski K.P.
      mRNA profiling for vaginal fluid and menstrual blood identification.
      Forensic Sci. Int. Genet. 2013; 7: 272-278
      ,
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ]; it is also suggested to be present in foreskin secretion [
      • van den Berge M.
      • Bhoelai B.
      • Harteveld J.
      • Matai A.
      • Sijen T.
      Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
      Forensic Sci. Int. Genet. 2016; 20: 119-129
      ]. The remaining two markers, MYOZ1 and CYP2B7P1, had better performance and showed a higher degree of co-expression with other mRNA vaginal mucosa markers.
      The scoring model for the detection of vaginal mucosa followed a ≥ 50 % threshold of detected markers in the RNA profiles as recommended by Lindenbergh et al. [
      • Lindenbergh A.
      • Maaskant P.
      • Sijen T.
      Implementation of RNA profiling in forensic casework.
      Forensic Sci. Int. Genet. 2013; 7: 159-166
      ]. If a sample had three valid RNA profiles at least five mRNA vaginal mucosa peaks had to be observed (Section 2.3.3). Sporadic detection, i.e., < 50 % of mRNA vaginal mucosa markers, were observed in 35 % of all the samples. Likewise 53 % of samples gave a sporadic detection for blood (Supplementary material 1, Table S3). Sporadic detection was considered to be unreliable as they are probably low-level components or spurious signals due to high cDNA input [
      • Lindenbergh A.
      • Maaskant P.
      • Sijen T.
      Implementation of RNA profiling in forensic casework.
      Forensic Sci. Int. Genet. 2013; 7: 159-166
      ]. Van den Berge et al. [
      • van den Berge M.
      • Ozcanhan G.
      • Zijlstra S.
      • Lindenbergh A.
      • Sijen T.
      Prevalence of human cell material: DNA and RNA profiling of public and private objects and after activity scenarios.
      Forensic Sci. Int. Genet. 2016; 21: 81-89
      ] observed sporadic peaks in 70 % of 164 private samples collected from individual’s skin or clothing when results of body fluids/cell types in the mRNA multiplexes were evaluated.
      The information used to evaluate the evidence in the Bayesian network analysis showed that a negative or positive vaginal mucosa test had minimal impact compared to the DNA results (RFU̅POI). An equivalent continuous model for reporting mRNA body fluids based on RFU values is problematic to achieve as the RNA quantification method is not human or body fluid specific, hence neither mixture proportions nor PCR dilution factor can be calculated. In addition, the RFU values of the mRNA vaginal mucosa markers, in our study, were similar for both intimate contact and non-intimate contact samples, except for the “0 h” intimate samples which tended to be higher.

      4.3 Bayesian network

      The Bayesian network analysis and the case examples illustrates how the findings in this study can be used to evaluate activity level propositions in casework. Highest recoveries and weights of evidence for a given log10RFU̅POIwere achieved from boxershorts, followed by penile swabs and fingernail swabs. These data can be considered to be used to help interpret other types of evidence, e.g. condoms, that involve transfer of vaginal mucosa. The probability of detecting a positive result for vaginal mucosa for intimate contact samples were similar across the three evidence types (Fig. 10), but in comparison to the RFU̅POI, the impact on the LR is low. The decay of RFU̅POI over time was more pronounced with fingernail swabs, hence the achievable LR over time is lower compared to penile swabs. Furthermore, the data analysis shows that boxershorts are an excellent source of evidential material, since it is not time dependent.

      4.4 Source vs activity level evaluations for mRNA models

      Source level evaluation of the evidence to assign the body fluid type was not attempted. If carried out, it is necessary for a court to definitively assign a body fluid result before proceeding to activity level, which includes direct vs. indirect transfer, and the effect of background. However, within our activity level framework, there is no requirement to definitively identify the body fluid; indeed, high LRs in support of the prosecution proposition are still achieved in the absence of evidence for a positive vaginal mucosa RNA result in the crime stain. Including body fluid tests undoubtedly adds value to the evaluation of the evidence but the evaluation is simplified because the source level interpretation is omitted. Imperfect (presumptive) mRNA tests, are well suited to direct incorporation into activity level calculations. As mRNA tests improve, so will the value of the evidence that is produced.

      4.5 The appeal of R v Weller

      An early example of evaluation using DNA quantity in relation to sexual assault, is provided in the appeal of R v Weller [

      R. v Weller, 2010. EWCA Crim 1085 (UK). <〈http://www.bailii.org/ew/cases/EWCA/Crim/2010/1085.html〉>).

      ]. The victim claimed that the defendant had sexually assaulted her, whereas the defendant claimed that only social contact occurred when he helped her to bed and touched her hair, when she became intoxicated at a party. The evidence was a DNA mixture underneath fingernails of the defendant's left hand – he was the major contributor and the victim was present (agreed by the court) as the minor contributor (a full profile was present). There was no attempt to identify vaginal mucosa in this case, hence the evidence rested solely upon the observation of the victim's DNA underneath fingernails of the defendant. The value of the evidence was not quantified using likelihood ratios, although it was described as providing strong support for the prosecution allegation. The appeal against the conviction was not allowed (i.e., the defendant's guilt was upheld). The case is described in detail in [

      R. v Weller, 2010. EWCA Crim 1085 (UK). <〈http://www.bailii.org/ew/cases/EWCA/Crim/2010/1085.html〉>).

      ] where propositions are listed, along with an example statement. The data provided in this paper now give the opportunity to carry out numerical evaluation of the evidence for this kind of case.

      4.6 Future research

      For future research we suggest that more mRNA markers should be examined which are present at high abundance in vaginal mucosa and perform well in primer multiplexes. However, it may be difficult to find highly specific markers for vaginal mucosa as epithelial cells have common functions, e.g. secretion and protection [
      • Hanson E.K.
      • Ballantyne J.
      Highly specific mRNA biomarkers for the identification of vaginal secretions in sexual assault investigations.
      Sci. Justice. 2013; 53: 14-22
      ]. The mRNA marker MUC4 should be considered for replacement as it did not perform well in this study. Activity level frameworks may become the preferred option to evaluate body fluid evidence, by-passing the need to evaluate the evidence at source level.

      CRediT authorship contribution statement

      Helen Johannessen: Conceptualization, Investigation, Formal analysis, Methodology, Data curation, Visualization, Validation, Writing – original draft. Peter Gill: Conceptualization, Methodology, Formal analysis, Data curation, Validation, Visualization, Supervision, Bayesian Networks, Writing - original draft, Writing - review & editing. Gnanagowry Shanthan: Investigation, Writing – review & editing. Ane Elida Fonneløp: Conceptualization, Formal analysis, Methodology, Data curation, Visualization, Validation, Writing – review & editing, Project administration, Supervision.

      Declaration of Competing Interest

      The authors declare that they have no known competing interest that could have influenced the work presented in this paper.

      Acknowledgements

      The project was funded by UiO: Life Science, Norway (grant ID 2018/10221). We would like to thank all the volunteers supplying samples for this project. In addition, we would like to thank Dr. Øyvind Bleka for his help with analysis based on EuroForMix and for his constructive comments on the manuscript.

      Appendix A. Supplementary material

      .
      .
      .
      .
      .
      .
      .

      References

        • Harbison S.A.
        • Fleming R.
        Forensic body fluid identification: state of the art.
        Res. Rep. Forensic Med. Sci. 2016; 6: 11-23
        View in Article
        • Virkler K.
        • Lednev I.K.
        Analysis of body fluids for forensic purposes: from laboratory testing to non-destructive rapid confirmatory identification at a crime scene.
        Forensic Sci. Int. 2009; 188: 1-17
        View in Article
        • Juusola J.
        • Ballantyne J.
        Messenger RNA profiling: a prototype method to supplant conventional methods for body fluid identification.
        Forensic Sci. Int. 2003; 135: 85-96
        View in Article
        • Haas C.
        • Hanson E.
        • Anjos M.J.
        • Bar W.
        • Banemann R.
        • Berti A.
        • Borges E.
        • Bouakaze C.
        • Carracedo A.
        • Carvalho M.
        • Castella V.
        • Choma A.
        • De Cock G.
        • Dotsch M.
        • Hoff-Olsen P.
        • Johansen P.
        • Kohlmeier F.
        • Lindenbergh P.A.
        • Ludes B.
        • Maronas O.
        • Moore D.
        • Morerod M.L.
        • Morling N.
        • Niederstatter H.
        • Noel F.
        • Parson W.
        • Patel G.
        • Popielarz C.
        • Salata E.
        • Schneider P.M.
        • Sijen T.
        • Sviezena B.
        • Turanska M.
        • Zatkalikova L.
        • Ballantyne J.
        RNA/DNA co-analysis from blood stains--results of a second collaborative EDNAP exercise.
        Forensic Sci. Int. Genet. 2012; 6: 70-80
        View in Article
        • Haas C.
        • Hanson E.
        • Anjos M.J.
        • Banemann R.
        • Berti A.
        • Borges E.
        • Carracedo A.
        • Carvalho M.
        • Courts C.
        • De Cock G.
        • Dotsch M.
        • Flynn S.
        • Gomes I.
        • Hollard C.
        • Hjort B.
        • Hoff-Olsen P.
        • Hribikova K.
        • Lindenbergh A.
        • Ludes B.
        • Maronas O.
        • McCallum N.
        • Moore D.
        • Morling N.
        • Niederstatter H.
        • Noel F.
        • Parson W.
        • Popielarz C.
        • Rapone C.
        • Roeder A.D.
        • Ruiz Y.
        • Sauer E.
        • Schneider P.M.
        • Sijen T.
        • Court D.S.
        • Sviezena B.
        • Turanska M.
        • Vidaki A.
        • Zatkalikova L.
        • Ballantyne J.
        RNA/DNA co-analysis from human saliva and semen stains--results of a third collaborative EDNAP exercise.
        Forensic Sci. Int. Genet. 2013; 7: 230-239
        View in Article
        • Haas C.
        • Hanson E.
        • Anjos M.J.
        • Ballantyne K.N.
        • Banemann R.
        • Bhoelai B.
        • Borges E.
        • Carvalho M.
        • Courts C.
        • De Cock G.
        • Drobnic K.
        • Dotsch M.
        • Fleming R.
        • Franchi C.
        • Gomes I.
        • Hadzic G.
        • Harbison S.A.
        • Harteveld J.
        • Hjort B.
        • Hollard C.
        • Hoff-Olsen P.
        • Huls C.
        • Keyser C.
        • Maronas O.
        • McCallum N.
        • Moore D.
        • Morling N.
        • Niederstatter H.
        • Noel F.
        • Parson W.
        • Phillips C.
        • Popielarz C.
        • Roeder A.D.
        • Salvaderi L.
        • Sauer E.
        • Schneider P.M.
        • Shanthan G.
        • Court D.S.
        • Turanska M.
        • van Oorschot R.A.
        • Vennemann M.
        • Vidaki A.
        • Zatkalikova L.
        • Ballantyne J.
        RNA/DNA co-analysis from human menstrual blood and vaginal secretion stains: results of a fourth and fifth collaborative EDNAP exercise.
        Forensic Sci. Int. Genet. 2014; 8: 203-212
        View in Article
        • van den Berge M.
        • Carracedo A.
        • Gomes I.
        • Graham E.A.
        • Haas C.
        • Hjort B.
        • Hoff-Olsen P.
        • Maronas O.
        • Mevag B.
        • Morling N.
        • Niederstatter H.
        • Parson W.
        • Schneider P.M.
        • Court D.S.
        • Vidaki A.
        • Sijen T.
        A collaborative European exercise on mRNA-based body fluid/skin typing and interpretation of DNA and RNA results.
        Forensic Sci. Int. Genet. 2014; 10: 40-48
        View in Article
        • Lindenbergh A.
        • de Pagter M.
        • Ramdayal G.
        • Visser M.
        • Zubakov D.
        • Kayser M.
        • Sijen T.
        A multiplex (m)RNA-profiling system for the forensic identification of body fluids and contact traces.
        Forensic Sci. Int. Genet. 2012; 6: 565-577
        View in Article
        • Albani P.P.
        • Fleming R.
        Developmental validation of an enhanced mRNA-based multiplex system for body fluid and cell type identification.
        Sci. Justice. 2019; 59: 217-227
        View in Article
        • Jakubowska J.
        • Maciejewska A.
        • Pawlowski R.
        • Bielawski K.P.
        mRNA profiling for vaginal fluid and menstrual blood identification.
        Forensic Sci. Int. Genet. 2013; 7: 272-278
        View in Article
        • Roeder A.D.
        • Haas C.
        mRNA profiling using a minimum of five mRNA markers per body fluid and a novel scoring method for body fluid identification.
        Int. J. Leg. Med. 2013; 127: 707-721
        View in Article
        • Hanson E.K.
        • Ballantyne J.
        Highly specific mRNA biomarkers for the identification of vaginal secretions in sexual assault investigations.
        Sci. Justice. 2013; 53: 14-22
        View in Article
        • Richard M.L.
        • Harper K.A.
        • Craig R.L.
        • Onorato A.J.
        • Robertson J.M.
        • Donfack J.
        Evaluation of mRNA marker specificity for the identification of five human body fluids by capillary electrophoresis.
        Forensic Sci. Int. Genet. 2012; 6: 452-460
        View in Article
        • Cossu C.
        • Germann U.
        • Kratzer A.
        • Bär W.
        • Haas C.
        How specific are the vaginal secretion mRNA-markers HBD1 and MUC4?.
        Forensic Sci. Int. Genet. Suppl. Ser. 2009; 2: 536-537
        View in Article
        • Blackman S.
        • Stafford-Allen B.
        • Hanson E.K.
        • Panasiuk M.
        • Brooker A.-L.
        • Rendell P.
        • Ballantyne J.
        • Wells S.
        Developmental validation of the ParaDNA® body fluid ID system – a rapid multiplex mRNA-profiling system for the forensic identification of body fluids.
        Forensic Sci. Int. Genet. 2018; 37: 151-161
        View in Article
        • van den Berge M.
        • Bhoelai B.
        • Harteveld J.
        • Matai A.
        • Sijen T.
        Advancing forensic RNA typing: on non-target secretions, a nasal mucosa marker, a differential co-extraction protocol and the sensitivity of DNA and RNA profiling.
        Forensic Sci. Int. Genet. 2016; 20: 119-129
        View in Article
        • van den Berge M.
        • Ozcanhan G.
        • Zijlstra S.
        • Lindenbergh A.
        • Sijen T.
        Prevalence of human cell material: DNA and RNA profiling of public and private objects and after activity scenarios.
        Forensic Sci. Int. Genet. 2016; 21: 81-89
        View in Article
        • Lacerenza D.
        • Aneli S.
        • Omedei M.
        • Gino S.
        • Pasino S.
        • Berchialla P.
        • Robino C.
        A molecular exploration of human DNA/RNA co-extracted from the palmar surface of the hands and fingers.
        Forensic Sci. Int. Genet. 2016; 22: 44-53
        View in Article
        • Cook O.
        • Dixon L.
        The prevalence of mixed DNA profiles in fingernail samples taken from individuals in the general population.
        Forensic Sci. Int. Genet. 2007; 1: 62-68
        View in Article
        • Malsom S.
        • Flanagan N.
        • McAlister C.
        • Dixon L.
        The prevalence of mixed DNA profiles in fingernail samples taken from couples who co-habit using autosomal and Y-STRs.
        Forensic Sci. Int. Genet. 2009; 3: 57-62
        View in Article
        • Matte M.
        • Williams L.
        • Frappier R.
        • Newman J.
        Prevalence and persistence of foreign DNA beneath fingernails.
        Forensic Sci. Int. Genet. 2012; 6: 236-243
        View in Article
        • Fonneløp A.E.
        • Johannessen H.
        • Heen G.
        • Molland K.
        • Gill P.
        A retrospective study on the transfer, persistence and recovery of sperm and epithelial cells in samples collected in sexual assault casework, Forensic Science.
        Forensic Sci. Int. Genet. 2019; 43
        View in Article
        • Bouzga M.M.
        • Dørum G.
        • Gundersen K.
        • Kohler P.
        • Hoff-Olsen P.
        • Fonneløp A.E.
        Is it possible to predict the origin of epithelial cells? – a comparison of secondary transfer of skin epithelial cells versus vaginal mucous membrane cells by direct contact.
        Sci. Justice. 2020; 60: 234-242
        View in Article
        • Cina S.J.
        • Collins K.A.
        • Pettenati M.J.
        • Fitts M.
        Isolation and identification of female DNA on postcoital penile swabs.
        Am. J. Forensic Med. Pathol. 2000; 21: 97-100
        View in Article
        • Farmen R.K.
        • Haukeli I.
        • Ruoff P.
        • Froyland E.S.
        Assessing the presence of female DNA on post-coital penile swabs: relevance to the investigation of sexual assault.
        J. Forensic Leg. Med. 2012; 19: 386-389
        View in Article
        • Kaarstad K.
        • Rohde M.
        • Larsen J.
        • Eriksen B.
        • Thomsen J.L.
        The detection of female DNA from the penis in sexual assault cases.
        J. Forensic Leg. Med. 2007; 14: 159-160
        View in Article
        • Gill P.
        • Hicks T.
        • Butler J.M.
        • Connolly E.
        • Gusmao L.
        • Kokshoorn B.
        • Morling N.
        • van Oorschot R.A.H.
        • Parson W.
        • Prinz M.
        • Schneider P.M.
        • Sijen T.
        • Taylor D.
        DNA commission of the International society for forensic genetics: assessing the value of forensic biological evidence – guidelines highlighting the importance of propositions. Part II: evaluation of biological traces considering activity level propositions.
        Forensic Sci. Int. Genet. 2020; 44102186
        View in Article
        • Wang S.
        • Shanthan G.
        • Bouzga M.M.
        • Dinh H.M.T.
        • Haas C.
        • Fonneløp A.E.
        Evaluating the performance of five up-to-date DNA/RNA coextraction methods for forensic application.
        Forensic Sci. Int. 2021; 328
        View in Article
        • van den Berge M.
        • Sijen T.
        A male and female RNA marker to infer sex in forensic analysis.
        Forensic Sci. Int. Genet. 2017; 26: 70-76
        View in Article
        • van den Berge M.
        • Sijen T.
        Extended specificity studies of mRNA assays used to infer human organ tissues and body fluids.
        Electrophoresis. 2017; 38: 3155-3160
        View in Article
        • Lindenbergh A.
        • Maaskant P.
        • Sijen T.
        Implementation of RNA profiling in forensic casework.
        Forensic Sci. Int. Genet. 2013; 7: 159-166
        View in Article
        • Bleka Ø.
        • Storvik G.
        • Gill P.
        EuroForMix: an open source software based on a continuous model to evaluate STR DNA profiles from a mixture of contributors with artefacts.
        Forensic Sci. Int. Genet. 2015; 21: 35-44
        View in Article

      33. ENFSI, E.N.F.S.I. guideline for evaluative reporting in forensic science, 2016.

        View in Article
        • Gill P.
        • Bleka Ø.
        • Fonneløp A.E.
        RFU derived LRs for activity level assignments using Bayesian Networks.
        Forensic Sci. Int. Genet. 2022; 56102608
        View in Article
        • Taylor D.
        • Biedermann A.
        • Hicks T.
        • Champod C.
        A template for constructing Bayesian networks in forensic biology cases when considering activity level propositions.
        Forensic Sci. Int. Genet. 2018; 33: 136-146
        View in Article
        • Biedermann A.
        • Taroni F.
        Bayesian networks for evaluating forensic DNA profiling evidence: a review and guide to literature.
        Forensic Sci. Int. Genet. 2012; 6: 147-157
        View in Article
        • Gill P.
        • Bleka Ø.
        • Roseth A.
        • Fonneløp A.E.
        An LR framework incorporating sensitivity analysis to model multiple direct and secondary transfer events on skin surface.
        Forensic Sci. Int. Genet. 2021; 53 (102509-102509)
        View in Article
        • Sijen T.
        • Harbison S.
        On the identification of body fluids and tissues: a crucial link in the investigation and solution of crime.
        Genes. 2021; 12: 1728
        View in Article
        • Muth C.
        • Oravecz Z.
        • Gabry J.
        User-friendly Bayesian regression modeling: a tutorial with rstanarm and shinystan.
        Tutor. Quant. Methods Psychol. 2018; 14: 99-119
        View in Article
        • Gill P.
        • Bleka O.
        • Hansson O.
        • Benschop C.
        • Haned H.
        Forensic Practioner’s Guide to the Interpretation of Complex DNA Profiles.
        Academic Press, 2020: 304-306
        View in Article
        • Sijen T.
        Molecular approaches for forensic cell type identification: On mRNA, miRNA, DNA methylation and microbial markers.
        Forensic Sci. Int. Genet. 2015; 18: 21-32
        View in Article

      42. R. v Weller, 2010. EWCA Crim 1085 (UK). <〈http://www.bailii.org/ew/cases/EWCA/Crim/2010/1085.html〉>).

        View in Article

      Article info

      Publication history

      Published online: July 21, 2022
      Accepted: July 19, 2022
      Received in revised form: July 18, 2022
      Received: February 7, 2022

      Identification

      DOI: https://doi.org/10.1016/j.fsigen.2022.102750

      Copyright

      © 2022 The Author(s). Published by Elsevier B.V.

      User license

      Creative Commons Attribution (CC BY 4.0) |
      How you can reuse Information Icon
      Creative Commons Attribution (CC BY 4.0)

      Permitted

      • Read, print & download
      • Redistribute or republish the final article
      • Text & data mine
      • Translate the article
      • Reuse portions or extracts from the article in other works
      • Sell or re-use for commercial purposes

      Elsevier's open access license policy

      ScienceDirect

      Access this article on ScienceDirect
      We use cookies to help provide and enhance our service and tailor content. To update your cookie settings, please visit the for this site.
      Copyright © 2023 Elsevier Inc. except certain content provided by third parties. The content on this site is intended for healthcare professionals.


      RELX