Distinguishing between Monozygotic Twins in Forensic DNA Analysis

Monozygotic Twins in Forensic

DNA Analysis

Literature Thesis MSc Forensic Science

Hannah German (12717274)

MSc Forensic Science

University of Amsterdam

Supervisor: Prof. Dr. A.D. Kloosterman

Examiner: Dr. M.J. Blom

October - December 2020

9187 words

1

1 ABSTRACT

In recent decades, the field of forensic DNA analysis has undergone incredible developments, making it the most valuable field of forensic science today. However, the method of STR analysis, which is the current standard practice used by forensic institutes around the world, is not able to distinguish between DNA profiles of monozygotic twins due to selection of specific markers in the DNA that are identical between them. This has lead to many cases involving twins in the past to remain unsolved or lead to a verdict based on merely circumstantial evidence. This thesis describes the novel advances that have been made in recent years in the field of DNA analysis that may be able to facilitate the distinction between DNA profiles of monozygotic twins in the future. First, a brief overview of the history of forensic DNA analysis is provided, from the moment it was first implemented in the 1980s to the methods that are currently used in casework. Then, the different possibilities for new methods are assessed, focusing on genetic and epigenetic approaches separately. The methods that may facilitate monozygotic twin analysis in a forensic context in the future in fact already exist, but when it comes to their application in the forensic field there are major roadblocks that researchers are facing. It may be too costly and labor-intensive to apply specific, time-consuming Whole Genome Sequencing methods for those few cases involving related people and twins, where simpler and quicker methods using just STRs or predetermined SNP and CpG markers suffice for most other cases. Furthermore, whereas methods such as Whole Genome Sequencing and Bisulfite Sequencing are already widely implemented in (epi)genomics research around the world, the forensic field requires more in-depth validation and standardization of the techniques before they can be used in casework. Therefore, although application of these new methods in the forensic field seems inevitably close, it may still be far away.

2

2 INTRODUCTION

On July 6th 2019, a 76 year-old woman was attacked and sexually assaulted while taking a walk out in the nature of Schipborg, Drenthe. With the help of witness statements, a 25-year old suspect was apprehended. A DNA profile was recovered from the clothing of the victim and, even though the suspect’s profile was not present in the database at the time, a match was yielded. This match turned out to be his identical twin brother. As the original suspect denies involvement in the crime and points the finger at his twin brother, who has a questionable alibi himself, investigators are still puzzled (Molenaar, 2020).

It is a generally accepted idea that monozygotic twins have identical DNA. However, this hypothesis has turned out to be old-fashioned. During their lifetime, starting from the moment the blastocyst splits, the DNA of each twin will undergo small changes, making them each unique. However, even though these differences are present, they cannot be detected with the standard forensic DNA analysis methods that are currently in use. Therefore, in the context of a forensic DNA analysis, monozygotic twins are still indistinguishable (Netherlands Forensic Institute, 2020). To be able to definitively tell which one of the twin brothers is the true donor of the DNA profile found on the victim’s clothing in the case described above, the Netherlands Forensic Institute (NFI) has to resort to other methods than the current PCR-CE STR profiling. Even though this will be a lengthy, costly and labor-intensive process, it is one of the last resorts for solid evidence to base a reliable verdict on in this case (Molenaar, 2020).

The Schipborg case is a unique one for the Netherlands, but this is not the first time DNA analysis of monozygotic twins plays a central role in a forensic investigation. A case that is well known on this topic is that of the brothers Dwayne and Dwight McNair, identical twins who both matched a DNA profile recovered from a rape scene. District Attorney David Deakin had been breaking his head over this case for years when he heard of new DNA techniques arising that could allegedly distinguish between the twin brothers. The test was performed at a cost of approximately 130.000 dollars and the results came in: the DNA matched Dwayne and not Dwight. A statistical analysis ruled that the DNA profile was approximately two billion times more likely to have been found under the hypothesis that the DNA trace came from Dwayne rather than from Dwight (Zimmer, 2019). This finding was wrongly interpreted by Deakin as a posterior probability: he in fact reported that it was two billion times more likely that the DNA originated from Dwayne than from Dwight. This is a classic example of a Prosecutor’s fallacy, which was never reported on. Eventually the outcome of the case was not influenced by this, as the DNA results were excluded following a motion filed by Dwayne McNair’s lawyers. According to them, the method was too new and not validated well enough to be applied in casework. This not only put the prosecutors back at square one, it also made prosecutors working on similar cases decide against using the technique. Ultimately, based on a witness statement, Dwayne McNair was still convicted and sentenced to 16 years in prison (Zimmer, 2019).

Twin-related cases do not always involve crime, but may also be of a different nature. A well-known example is the paternity case of Holly Marie Adams vs. Richard and Raymon Miller in 2004. Holly had sexual intercourse with both twin brothers Richard and Raymon around the same time and fell pregnant. After filing for a paternity test, the court could only rule that the probabilities of either brother being the father of the child were equal, as better tests did not exist at the time. Therefore, they ultimately had to resort to less reliable circumstantial evidence to rule that Raymon Miller is the father of the child (Missouri Court of Appeals, 2007). In this day and age, where we are starting to unravel even the smallest of details of our genome, it seems almost out of place that cases like those described above still remain unsolved. Whereas standard STR profiling looks at parts of the DNA that are generally the same between monozygotic twins, the key to distinguishing between them lies in very sparse,

3 slight changes present in their DNA. These may occur at random, but are also dependent on the environment and conditions they are exposed to during their lifetime, which are often different between the two. Furthermore, they may not only occur as mutations in the DNA (genetic), but also in a secondary layer as dynamic modifications of the DNA (epigenetic). If methods can be employed that allow us to detect these small differences, there will be a way to distinguish between monozygotic twins in forensic cases based on their DNA. These methods in fact already exist and are widely employed in many fields of research, but there are many roadblocks that have prevented them from being employed in a forensic context. For instance, they often require relatively expensive machinery and are not standardized or validated yet for forensic use.

This thesis will provide a review on which methods currently exist that are able to distinguish between monozygotic twins based on their DNA and how they are already able, or will in the future be able, to be applied in forensic casework. Many reviews have already been written on these methods separately, but a combined review on their application in the forensic field is still lacking. Firstly, a detailed overview will be given of how forensic DNA analysis has evolved, from the point it was first implemented in forensic practice in the 1980s to the methods that are currently employed in most cases. A critical assessment will be done of the current methods in the context of monozygotic twin analysis. Secondly, possible new methods and their applicability in the forensic field will be discussed. Attention will be given to their strengths and weaknesses, and where they might need alterations to be able to be used in actual casework. Lastly, a discussion of the relevance in casework and opportunities for future research is provided, from which several conclusions will be drawn.

3 DEVELOPMENT OF DNA ANALYSIS

In this section, a detailed overview of the development of forensic DNA analysis will be provided, starting from the moment it was first implemented in casework to the method of PCR-CE STR profiling that is currently widely used. The latter will be critically assessed on its effectiveness in twin cases, utilizing the case examples discussed in the introduction.

3.1 A

B

H

The forensic DNA analysis era started in the 1980s with the development of a method called Restriction Fragment Length Polymorphism (RFLP), which used restriction enzymes to cut out certain fragments of DNA, which were subsequently analyzed using X-ray based gel electrophoresis (Butler, 2009; Jeffreys et al., 1985). This method was based on a discovery made by Dr. Alec Jeffreys, who found repeated DNA sequences in the non-coding region of the human genome. The number of repeats seemed to differ between individuals, which gave them the potential to be used for the construction of an individual’s unique DNA profile (Jeffreys et al., 1985). Accordingly, Jeffreys named these repeated units VNTRs (Variable Number of Tandem Repeats). Although a breakthrough invention at the time, the RFLP method was far from ideal. Dependent on the restriction enzymes used, fragments were often large, making their migration through a gel very slow. On top of this, interpretations all had to be performed manually, making the entire process usually take somewhere between six and eight weeks. As a result of this, several laboratories decided to start using different restriction enzymes which would cut shorter fragments, making comparison of profiles produced by different laboratories virtually impossible. Lastly, only large quantities of intact DNA could be analyzed with the RLFP method, as the current standard method of PCR (Polymerase Chain Reaction) to amplify DNA fragments was not employed in the forensic field yet (Butler, 2009).

4 This is not a surprising fact, as PCR had only just been developed when RLFP made its way into forensic practice. However, it would later become one of the biggest advances in the forensic DNA analysis field. PCR is an enzymatic reaction that allows for the production of millions of copies of a single DNA fragment of interest in a matter of hours (Mullis et al., 1986). For forensic DNA analysis, this meant that samples of low DNA quantity were no longer an issue, as they could easily be amplified. Furthermore, it enabled the analysis of degraded or low-quality DNA samples, something that had hindered forensic investigations in the past (Butler, 2009). Therefore, it did not take long before RFLP and PCR were combined to be able to get feasible results from a wide range of forensic DNA samples (Jeffreys et al., 1988).

3.2 C

P

In the late 1990s, the shift to Short Tandem Repeats (STRs) was made, which are still the standard markers currently used. These repeats are generally only 3-5 bases long, making the total fragments to be analyzed much shorter (Caskey & Edwards, 1994). Furthermore, the X-ray based gel electrophoresis was replaced by capillary electrophoresis, a much faster and more automated way of detecting DNA fragments without having to cast an entire gel. In 1996, the first commercial multiplex STR kits became available, which were first implemented in the forensic labs by the year 2000. Multiplex STR kits allow for simultaneous PCR amplification and analysis of several STRs at once, greatly reducing the reaction time and enabling automation and standardization of DNA analysis methods (Butler, 2009). Each amplicon is given a different fluorescent label that is automatically detected in the capillary electrophoresis phase of the analysis. These developments created a more straightforward analysis procedure that can be performed by trained lab technicians as well as easier comparison between profiles and more robust results.

Today, a wide variety of standard commercial STR typing kits exist, allowing for a relatively fast and foolproof DNA analysis (Westen et al.,2014). The strength of this method mainly lies in the fact that only 13 STRs need to be analyzed to get a probability of less than 1 in a trillion of two unrelated people having exactly the same profile. However, in practice this number is rarely reached. Even though STR profiling is the current standard in forensic practice, it still has some drawbacks. Interpreting an STR profile usually means having to deal with several artifacts such as stutter peaks, off-ladder alleles and partial profiles due to drop-out of alleles. Therefore, the evidential strength resulting from it is heavily dependent on the quality of the analysis and of the DNA obtained. Furthermore, the analysis gets more complicated when dealing with DNA mixtures of several people, partial profiles or related people (Butler, 2009). Even though STRs are not in the coding region of the DNA, they still inherit like different alleles of coding genes would, following Mendelian genetics. For each locus, each individual inherits one allele from the father and one from the mother. This means that individuals that are related have a higher probability of having similar STR profiles than unrelated individuals would. That brings us back to the monozygotic twins, which may be the most extreme case of related people one can think of. As they are derived from one single zygote, all their STR alleles are the same. This renders STR profiling on twins useless, as their profiles would come out identical. One could imagine that they may have point mutations between them, which have randomly occurred during their lifetime. However, STR profiling is not able to detect this as STR alleles are separated based on their size and not based on their sequence. A single point mutation would not make a difference in a fragment’s size and therefore would yield an identical profile. The exception would be if a mutation is present in the primer binding site of a DNA fragment, causing the primer to bind with less affinity, hampering the amplification. This could lead to allele-dropout, which is still not very informative and could be interpreted as a mere artifact.

5 Now referring back to the Adams vs. Miller case, it makes perfect sense that the court could conclude no more than there being an equal probability of 99.9% that either of the men were the father. Both of their STR profiles were consistent with being the father of the child, but they were completely identical and thus neither could be ruled out based on this evidence (Missouri Court of Appeals, 2007). For the distinction to be made, more sophisticated techniques would have to be employed. Cases like this one among many others were thus the ones that made forensic experts speculate if there would be a way to make a more accurate distinction, if not a distinction at all, possible between twins and other closely related people while still maintaining the thorough validation and standardization associated with the current PCR -CE method.

Following this, for a while, the idea was raised that there must be some STRs present that differed between monozygotic twins that just simply had not been discovered yet. This would be an ideal situation for the forensic DNA field, as these STRs could be readily implemented in the already existing, optimized and validated STR profiling technique used in most forensic cases. Furthermore, this may also aid in distinguishing between people who are merely siblings or parent-child. However, this idea was slowly let go after analysis of thousands of STRs did not provide any feasible results (Butler, 2014). There have been some somatic mutations found in STRs between twins, but this phenomenon is extremely rare and thus not feasible for use in a forensic context (Wang et al., 2015). From this point, it was clear that more advanced methods would have to be implemented to resolve this issue.

4 NOVEL APPROACHES FOR DNA ANALYSIS

As it is established that the current methods in place for forensic DNA analysis are not sufficient to distinguish between monozygotic twins, there is now opportunity to look for possible methods that could. In this section, two possible approaches are provided: a genetic approach based on germline mutations in the DNA itself that have occurred after splitting of the zygote, and an epigenetic approach based on modifications of the DNA due to the environment and internal factors that are unique to each twin. A critical assessment of both of these approaches will be provided, highlighting not only their strengths but also the points where they may still be lacking when it comes to a forensic application.

4.1 A

G

A

The first and perhaps most obvious approach that can be taken towards twin DNA analysis is looking more closely at their genetic differences. Though very small in number, these differences have been shown to exist (Vogt et al., 2011; Weber-Lehman et al., 2014). Advantageously, genetic differences between twins have already been widely studied across many different areas due to their exceptional nature. This includes research into (psychiatric) diseases, heritability and gene-environment interactions (Dal et al., 2014; Nishioka et al., 2018). With this research, the longstanding hypothesis that twins are genetically identical was quickly disproven. It was for example shown that twins may possess Copy Number Variations (CNVs) in several genes, a result of extreme somatic mosaicism, making one twin more susceptible to certain diseases than the other (Bruder et al., 2008). This has opened a whole new research area into monozygotic twin pairs of which one has a disease or condition and the other does not.

6 Although still in an early stage at the time and not yet applied to many fields other than medical research, these novel findings inspired researchers in 2012 to conduct a simple thought experiment, suggesting the use of simple genetic differences between twins in paternity disputes such as that of Hollie Marie Adams described earlier (Krawczak et al., 2012). The thought process went as follows: In 98% of twin pregnancies, the zygote splits after having undergone less than or exactly 7 divisions. Mutations that have occurred during this time are present in both twins and therefore have no relevance in this case. However, after splitting of the zygote and before formation of the primordial germ cells, the independent embryos will have undergone another 14 cell divisions. After these divisions, the ca. 5000 primordial germ cells will then undergo about 365 divisions before they enter spermatogenesis in late puberty and become sperm cells. Based on the fact that there are 14 divisions before formation of primordial germ cells and 365 afterwards, the researchers postulated that there is at least a probability of 14/(14+365) = 3.7% that a heritable mutation arises before formation of the primordial germ cells in the embryonic gonads. On top of this, it is estimated that 48 mutations will occur per haploid genome per generation. The researchers assumed this occurs according to a Poisson distribution with λ ≥ 0.037 x 48 = 1.78. This gives us a probability of 1 − 𝑒−λ₌ 0.83. Based on these calculations, Krawczak et al. therefore postulated that there is at least an 83% probability that a monozygotic twin carries one or more germline mutations, detectable in a semen sample, that his twin brother does not have. This germline mutation is subsequently also detectable in the offspring of the twin carrying the mutation. This calculation was later confirmed in a proof of principle study where semen samples of monozygotic twins were taken, as well as a blood sample of the child of one of the twins (Weber-Lehmann et al., 2014). Ultra-deep Next Generation Sequencing (NGS) of the whole genome followed by confirmatory targeted Sanger sequencing revealed five Single Nucleotide Polymorphisms (SNPs) unique to father and child on chromosomes 4, 6, 11, 14 and 15. This finding is very striking, as it proves not only that mutations in twins arise after splitting of the blastocyst, but also that they are carried on into germline tissue and, as they later proved through buccal swabs, into somatic tissue as well. This could be used not only to distinguish between twins in cases where a DNA sample from the crime scene is available, but also for paternity testing on donated DNA samples.

Weber-Lehmann et al. describe looking for single SNPs in the lengthy DNA sequence as ‘looking for a needle in a haystack’. According to them, what made their search succeed, unlike others in the past who failed (Kunio et al, 2013; Baranzini et al., 2010), is their high DNA coverage of approximately 90-fold for the twins and 50-fold for the child. Through this, they were able to confidently distinguish actual SNPs from sequencing artifacts. Although these findings are very promising, in order for this to be applied in casework there must be a robust way for reporting this evidence in the form of a likelihood ratio (LR). For this purpose, Krawzcak et al. followed up with the development of a mathematical model in 2018, based on the number of DNA sequencing reads yielded. First, STR profiling is used to rule out any other possible donors than the twins in question, after which the model can be applied using NGS as shown by Weber-Lehmann et al. in their proof of principle study. They also showed that this model can be applied to several case scenarios. Initially they used real genetic data and data from real-life casework, but this was met with some criticism from peer-reviewers and editors (Copenhaver et al., 2018). In their opinion, there is a substantial risk of identifying the subjects through their DNA and compromising the anonymity of the study. Additionally, the study did not apply informed consent to their participants, but merely informed them of the purpose of the study. Therefore, they agreed that no genetic data or data from forensic casework would be included in this publication. One can imagine that this problem could also arise in real casework when ultra-deep Whole Genome Sequencing is applied. The currently used STRs are in the non-coding region of the DNA, therefore they do not contain information about

7 phenotype or disease susceptibility that is currently known. However, when the whole genome of a person is sequenced and possibility saved in a database, this could give rise to privacy concerns that have to be taken into account.

Although studies have shown that high-coverage sequencing unveils SNPs between monozygotic twins, and a model has been proposed to express this in the form of an LR, it is still the question whether forensic DNA laboratories will have the resources to apply Whole Genome Sequencing with such high coverage to their samples. After all, to be able to be applied within a forensic framework, the approach would have to be immensely sensitive and precise. The only mutations that can be used are those that arise before spermatogenesis but after separation of the morula, which is an extremely narrow window in time. So narrow even, that in 20% of cases no mutations would occur at all in this stage (Krawczak et al., 2012). Furthermore, when the mutations are present, they need to be able to be detected in the immensely large DNA sequence. On the other hand however, the test is very definitive. Only one single mutation would be needed to distinguish between the twins and exclude one of the two, if the confidence level associated with the method is high enough. Furthermore, this study only focused on heritable mutations that were also found in the child of one of the twins , as their main focus was on paternity disputes. They did find more SNPs when comparing the twins, but these were excluded as they were not found in the child (Weber-Lehmann et al., 2014). One can imagine, however, that in criminal cases where kinship analysis is not involved these could also be of evidential value. As mentioned previously, they also found that their detected SNPs were not only present in semen, but also in buccal mucosa. This would provide a manner of sampling in forensic casework that is less invasive and more ethical than taking semen samples, while obtaining the same results.

In a very recent follow-up publication, Rolf & Krawczak summarized six forensic cases from 2014 to 2019 where their proposed Whole Genome Sequencing method was applied to find SNPs between monozygotic male twins, one of which was the proof-of-principle mock case performed by Weber-Lehmann et al. The other five consist of four kinship cases and one criminal case. The latter was very successful, as 9 SNPs were found of which two were present in the crime scene evidence. This lead to a likelihood ratio of 1.24x104_{, ultimately solving the} case. The four kinship cased revealed SNP numbers ranging from 0 to 11, but only 0-2 SNPs were found in the offspring as well. Likelihood ratios in these cases ranged from 6.1 to 1225.5, providing weaker evidence but still being able to solve the case. In one of the four kinship cases, zero SNPs were found and a likelihood ratio could not be established. The fact that the majority of these cases, which could not be solved previously with other DNA analysis methods, have now been solved using Whole Genome Sequencing is already a very promising development and paves the way for more forensic laboratories around the world to employ this method. As the authors acknowledge however, the complications still lie in the validation of the method for wide use in the forensic field, for which systematic studies have to be set up (Rolf & Krawczak, 2020).

As both the thought experiment and the proof of principle study by Weber -Lehmann et al. suggest, finding de novo mutations between monozygotic twins cannot be done with PCR-CE alone, though allele-specific PCR can be conducted as a second step to screen the DNA found on a crime scene in a targeted manner after detecting SNPs in a reference sample. For this purpose, a PCR assay specific for the mutations found has to be set up. As it is not guaranteed that different monozygotic twin pairs will share SNPs between them, this may be a challenge in forensic casework when it comes to high-throughput and automated analysis as well as databasing. As a first step however, to find germline mutations, high-coverage Whole Genome Sequencing has to be applied on reference samples of the twins. Whole Genome Sequencing is generally done using a novel set of sequencing methods that have all been developed and

8 applied in research in recent years, therefore also termed Next Generation Sequencing (NGS) methods. The following subsection will go deeper into the specifics of NGS, its advantages and drawbacks and what it can provide for the forensic field.

4.1.1 Next Generation Sequencing (NGS)

Next Generation Sequencing (NGS) is an umbrella term for various methods that are able to sequence large stretches of DNA in parallel, therefore also termed Massively Parallel Sequencing (MPS). It was first implemented in 2005, following the ‘first generation’ of laboratory sequencing (Behjati & Tarpey, 2013). This generation was mainly dominated by dideoxy sequencing, also termed Sanger sequencing after its developer Fred Sanger (Sanger et al., 1977). It was the very first and only feasible sequencing method at the time and for that reason widely employed, but it had its drawbacks. Not only was it very time-consuming and labor-intensive, it also required radioactive materials for detection of the sequence in the early stages. Later, it was combined with Capillary Electrophoresis (CE) for a more automated detection. Although Sanger sequencing was the main method making the Human Genome Project in 2001 possible, there was still a high demand for less costly, high throughput sequencing techniques (Schuster, 2008). This prompted the development of NGS in the early 2000s, its main users being the fields of evolutionary biology and oncology. NGS is mainly revolutionary in its ability to produce millions of reads in picoliter volumes, taking a much shorter time than Sanger sequencing. The reads can subsequently be stitched together computationally, enabling the sequencing of an entire genome in less than a day. Unlike Sanger, where the DNA is first synthesized and then detected using capillary electrophoresis, in NGS the sequence is directly detected during synthesis. This greatly reduces reaction time and complexity (Behjati & Tarpey, 2013).

The first NGS method on the market was 454/Roche, which makes use of pyrosequencing. During DNA synthesis, pyrophosphate is released which in Roche sequencing is used by the enzyme luciferase to produce a flash of light. By washing a different nucleotide over the DNA fragments each time, the incorporation of that specific nucleotide is measured by a detector and the sequence is automatically read out. After the Human Genome Project, which took 13 years at a cost of 2700 million dollars, Roche sequencing was able to re-sequence the human genome in just five months for only 1.5 million dollars (Schuster, 2008). After Roche was developed, several other methods quickly followed. A second widely used NGS method is Illumina/HiSeq, which uses sequencing by synthesis (SBS). First, bridge amplification in small wells is used to produce clusters of identical DNA fragments. Then, the fragments are linearized and a synthesis reaction is initiated with the addition of terminator nucleotides (ddATP, ddGTP, ddCTP and ddTTP). These nucleotides contain cleavable fluorescent groups and blocking groups in order to prevent incorporation of several nucleotides at a time. Like for Roche, the fluorescent signals are automatically detected. Afterwards, the blocking groups are cleaved off and the next nucleotide can be incorporated (Metzker, 2010).

In 2012, a review study was conducted comparing Next Generation Sequencing to dideoxy sequencing in terms of throughput, quality and efficiency (Liu et al., 2012). At the time, Sanger was favored by many scientists in terms of read length. NGS methods generally produced shorter reads, whereas Sanger can produce reads as long as 900 bp at a time. However, in recent years, NGS methods have been improved in such a way that they are expected to soon be able to reach the same read lengths as Sanger sequencing (Schuster, 2008). Furthermore, NGS methods generally have a higher accuracy, a higher throughput and a much lower cost than dideoxy sequencing. At the time, it was estimated by Liu et al. that Sanger sequencing

9 would cost 2400 dollars per million bases, compared to only 7 cents for Illumina/HiSeq. One can only imagine that in this day, this cost will have been reduced even further.

4.1.2 Next Generation Sequencing in a forensic context

These new sequencing methods all seem very promising and have already shown to be very valuable in the field of genetic research, but they have not been widely implemented in the forensic field yet. When it comes to a forensic application, it is crucial that DNA analysis is performed in a manner that is as accurate and reproducible as possible, while maintaining low costs, time efficiency and ease of use. Forensic DNA experts often have to deal with a high caseload while still being expected to produce reliable results that can be efficiently communicated to laymen and used in a court of law. Considering this, NGS has two major advantages over the current methods. Firstly, it is very time-efficient. Thousands to millions of reactions can occur in parallel, generating sequence reads very rapidly. This is especially important regarding the high caseload that forensic DNA laboratories may deal with as well as the time pressure of solving a high-profile case. Secondly, the use of NGS eliminates the need for electrophoresis to read out the sequence, as the output is directly registered during the synthesis reaction (Børsting & Morling, 2015). This also reduces the time it takes for the analysis to be performed as well as making the whole process more straightforward, which will not only benefit the laboratories but also the court when it comes to interpretation. There is no need to do a separate PCR and CE reaction, as the whole analysis can be done at once with reads being produced automatically.

The latter however can be seen as both a blessing and a curse. It is very efficient that reads are being produced quickly and automatically, but this is also a major roadblock that is currently being faced when it comes to Next Generation Sequencing being widely employed in forensics. Sequencing output generates gigabytes of data, which requires large storage and extensive data analysis. Furthermore, in contrast to PCR-CE, there is no way for the sequencing output to be assessed manually. The analysis therefore completely relies on the software and its error rate, as little human judgement is involved in the validity of the sequencing process. Therefore, scientists have been hesitant to employ the method in forensics as especially in this field, the software must be as highly validated and accurate as possible, which is not yet the case (Børsting & Morling, 2015). For example, it is known that some loci are more prone to mutation or stutter than others and thus they must be evaluated differently. This is especially important for forensics when calculating the rarity of a DNA profile and evaluating possible sequencing artifacts. However, this possibility is not yet sufficiently implemented in NGS methods for use the forensic genetics field. Furthermore, forensic laboratories are very dependent on differing in-house validation studies and accreditations, which need to be adhered to in the software they employ. When following the ISFG recommendations on DNA analysis, the ‘black box method’ that NGS currently comprises will likely not suffice in most forensic laboratories (Børsting & Morling, 2015; Gjertson et al., 2007).

When employing high-throughput sequencing in forensic twin cases, the most ideal approach would be to use shotgun sequencing. This is a method where the DNA is sequenced without selecting specific targets beforehand, yielding a lot of sequence information (Motahari et al., 2013). In monozygotic twin cases, where we may be looking for just several small differences in the immensely large genome, this would be of great help. There are however some drawbacks: for shotgun sequencing, microgram amounts of DNA are necessary as there is no targeted amplification step beforehand. In many forensic samples, these amounts are not obtained and they are thus not suitable for this approach unless whole genome amplification is performed. Furthermore, shotgun sequencing is not very reproducible as the stretches that

10 are sequenced are random and not determined beforehand (Børsting & Morling, 2015). Not only would this hinder the comparison between a sample found on a crime scene and a reference sample, it would also make databasing, standardization and quality control very difficult. Comparison of samples would then require sequencing of the entire genome to find parallel sequences. However, in cases that do not involve monozygotic twins or related people, sequencing the entire genome rather than just several markers with high discriminatory power is often excessive and too time-consuming. It would therefore make employing this method for just a few cases among many very costly and inefficient. On the other hand, Whole Genome Sequencing has become increasingly more cheap over the last years and has been shown by Weber-Lehmann et al. to be capable of finding SNPs between twins and resolve a hypothetical paternity dispute. It might therefore be possible in the future to employ this method for all forensic DNA samples and not just those requiring more extensive analysis, if the costs have dropped sufficiently and the software has been validated thoroughly for forensic use.

The alternative is a method called targeted sequencing, where regions selected beforehand are first amplified with PCR and then sequenced (Voelkerding et al., 2009). Due to the amplification step, this method only requires nanogram amounts of DNA, a number that is way more realistic when it comes to forensic casework. In recent years, Illumina has already brought a forensic DNA analysis kit on the market using this approach, called the ForenSeq DNA Signature Prep Kit (Illumina, San Diego, CA, USA). Their goal is to completely replace the current method of PCR-CE by PCR-NGS using this kit, which is able to detect over 200 markers, including autosomal, X-chromosomal and Y-chromosomal STRs as well as several SNPs giving information on identity, phenotype and ancestry. Several evaluations of this kit performed in the years 2017-2020 showed that ForenSeq is indeed a very promising method (Silvia et al, 2017; Köcher et al., 2018; Sharma et al., 2020). It only requires nanogram amounts of DNA to produce a full profile, while minimizing drop-out and being able to resolve mixtures. On the other hand, bringing new kits and methods into the forensic field does bring some difficulties. There is for example no consensus yet on how all this new information should be databased, as the current database does not contain all the STRs and SNPs that ForenSeq uncovers. Furthermore, as mentioned previously, new methods that are implemented in forensic analysis should first undergo in-house validation and accreditation before they can be confidently implemented in casework.

4.1.3 Concluding remarks

So what does all of this mean for forensic DNA analysis in the future? We can firstly conclude that high-throughput sequencing methods comprise a promising development in the field of DNA analysis in terms of efficiency and detail of analysis. It has been shown that NGS is capable of finding even the fewest of genetic differences between monozygotic twins as long as the coverage is sufficient, something that PCR-CE is not able to do. At this point, the implementation of NGS in the forensic field in the future through kits such as ForenSeq is almost unavoidable. However, there is still a long way to go when it comes to validation and advances in the software used for high-throughput sequencing. For application in casework, the software must be able to deal with complexities such as high ratio mixtures and differential evaluation of loci while still giving the most robust and trustworthy results possible. As genotype calls cannot be assessed manually when using this method, it has to be optimized and validated greatly before sufficient trust is won in the forensic community for implementation in casework. Furthermore, a general consensus has to be reached about how to deal with the big data NGS yields and how to approach databasing.

When it comes to twin cases, it is highly likely that we will have to go one step further and apply Whole Genome Sequencing to DNA samples rather than simply a targeted sequencing method using a standardized kit, which would suffice for regular DNA samples where the discriminatory

11 power does not have to be as high. With Whole Genome Sequencing becoming increasingly more cheap over the past years, this is not an unrealistic option. It has even already been applied in several criminal and kinship cases. However, when using this method, it is more difficult to make standardized kits and set up databases for comparison of profiles from different laboratories. Furthermore, a lot more data is yielded than for a targeted approach. Therefore, the issues with NGS highlighted before become amplified. This would mean that for Whole Genome Sequencing to be implemented in the forensic DNA field, it may only be feasible for use in those cases where yielding that much data is necessary and unavoidable. For other cases, a targeted sequencing method such as ForenSeq would be more realistic. On the other hand, with Whole Genome Sequencing becoming more cheap and automated over the past years, now only requiring straightforward benchtop sequencing machinery, it may be possible to widely implement it in the forensic DNA analysis field in the near future. This would reduce the costs of the analyses, as they can now be done in bulk rather than only a few cases among many, and would facilitate the opportunity for automation and standardization.

4.2 A

A

As an alternative to looking at genetic differences between monozygotic twins for DNA comparison, additional possibilities lie in the comparison of their epigenetic differences. As discussed in the previous section, rare genetic differences between twins may be the cause of different susceptibility to diseases or difference in general phenotypic features. Another explanation for these phenomena could be a variation in their epigenetic patterns. This is known to arise as a result of several factors, such as smoking and exercise habits, diet and aging. One can imagine that even though twins may be genetically almost identical from birth, they are exposed to different environmental factors during their lifetime and thus show a different epigenetic makeup. This hypothesis has already been confirmed in disease research, where it was for instance recently found that monozygotic twins who are discordant for congenital heart diseases such as Tetralogy of Fallot (TOF) show differential methylation patterns in regulatory sequences of several cardiac genes, despite having the same genetic makeup (Grunert et al., 2020). In a more general study by Fraga et al. in 2005, a large cohort of monozygotic twins of different ages was used to analyze patterns of DNA methylation as well as histone acetylation. They found that at a young age, the methylation and acetylation patterns of monozygotic twins are very similar. However, as the twins age, their epigenetic patterns become increasingly different, allowing for their distinction. These findings may bring new opportunities to the forensic field, where distinguishing between monozygotic twins on a genetic level can prove to be difficult and costly. In this section, the developments and techniques used in studying epigenetics will first be discussed, after which an assessment will be made of the opportunities it may bring to the forensic field in general and the comparison of monozygotic twin DNA. It will also be compared to the previously discussed methods using SNPs. As Fraga et al. showed in their study that twins differ significantly in DNA methylation, this DNA modification will be the main focus.

4.2.1 Techniques and developments in epigenetics

The term epigenetics is generally defined as heritable modifications of the DNA that do not affect the DNA sequence, but can be modified by environmental and internal factors. These modifications play a crucial role in gene expression and silencing, through regulating transcription and modifying the structure of chromatin. An important modification is DNA methylation, which usually occurs on cytosine nucleotides and is associated with transcriptional repression and gene silencing (Alberts et al., 2018). As different people, including twins, undergo different environmental and internal influences as a result of their

12 lifestyle, their DNA methylation pattern is expected to be different. This is very beneficial when trying to compare related people with a very similar DNA profile, but can also prove to be more difficult as epigenetic marks are very dynamic and may quickly change.

As the field of epigenetics has expanded in the past years, several different methods have been employed to detect DNA methylation. The most widely used method of these is bisulfite conversion, in which the chemical sodium bisulfite is added to single strands of DNA, which causes the conversion of unmethylated cytosines to uracil by deamination. Cytosines that are methylated, on the other hand, are protected from conversion. As a next step, the DNA sequence can be analyzed with either PCR or Next Generation Sequencing to assess which cytosines have been converted (Ogino et al., 2006). An advantage of using NGS for this purpose is that the converted cytosines can be treated as a SNP, so a method similar to that used in forensic genetics can be employed here. Methylation calling then happens simultaneously with mapping the reads into the reference sequence and the methylation detection does not have to occur in a targeted fashion. This is especially useful for monozygotic twin analysis as, similarly to the genetic approach described previously, we are looking for very small differences in a large genome. However, when reliable recurring markers are discovered in the future, it will also be possible to only sequence specific regions rather than the entire genome. The latter is mainly where the forensic interests lie, as genome-wide Bisulfite Sequencing is more costly. Aside from bisulfite conversion, there are alternative methods that can be combined with NGS to sequence DNA methylation, such as MeDIP , which uses specific antibodies for immunoprecipitation of methylated DNA.

For a forensic application as well as well as in other types of research, discovery of candidate age-associated, tissue-associated or lifestyle-associated epigenetic markers is an important component (Vidaki & Kayser, 2018). The main method used for this is a DNA methylation microarray, for which Illumina has already developed a standardized assay. This method works with immobilized probes to which methylated DNA can bind, allowing for the discovery of a large number of CpG sites at once. Furthermore, the forensic field can largely benefit from marker discovery in the biomedical field, as more and more is unraveled about how different methylation markers may affect susceptibility for diseases and cause discordance in monozygotic twins.

4.2.2 Forensic application of epigenetics

In the forensic field, epigenetics have already been widely used in several different areas, such as tissue typing of biological samples through tissue-specific epigenetic patterns and age prediction of unknown donors. Recently, as a result of the previously described difficulties associated with merely genetic approaches, the possibility of using epigenetics to distinguish between monozygotic twins has emerged. To test the feasibility of such an approach, a study was conducted in 2011, which used blood samples from 22 adult twin pairs to analyze their DNA methylation patterns and find possible markers for distinction (Li et al., 2011). Using bisulfite conversion followed by Illumina BeadChip analysis, they found significant methylation differences in 377 CpG sites, 92 of which were present in all subjects. Even though it was already known that DNA methylation on CpG sites is relatively unstable, this initial study into using DNA methylation for monozygotic twin distinction inspired many other studies. One of those was conducted in 2017 and used a similar method to find an additional set of six recurring markers with a differential methylation frequency of approximately 33% (Park et al., 2017). Even though this percentage is relatively low, it could still function as a preliminary test to decrease the number of expensive genome-wide searches that are necessary to detect genetic or epigenetic differences between monozygotic twins. Furthermore, it forms the basis for research into more such markers with a higher differential methylation percentage. Finding epigenetic markers for quick screening is especially important for a forensic application above

13 all others, as in this field it is crucial that analyses can be performed in a quick, cheap and automated manner. When a set of reliable CpG markers can be employed in forensics for detection of DNA methylation, this will enable the production of standard kits and eliminate the need for time-consuming and costly genome-wide screens. However, the markers used do need to be reliable to such an extent that their use can be justified for use in a court of law. The drawback of an epigenetic approach in forensics is that it is not compatible with all types of reference and trace samples found in forensic casework, as the DNA needs to be of relatively high quantity and quality to perform a genome-wide analysis. Furthermore, when there is a significant time difference between collection of different samples, such as trace and reference samples, the epigenetic patterns could have changed. The applicability of epigenetic analysis to trace and reference samples was studied by Viraki et al. in 2017. They collected whole blood from 10 female monozygotic twin pairs as a reference sample and simultaneously used this blood to make small dried blood spots on cotton material to act as a ‘trace’ sample. After isolating the DNA, they performed bisulfite conversion followed by either Illumina BeadChip analysis for genome-wide methylation profiling, or PCR. This lead to the discovery of several potential candidate CpG markers, just like the previous studies did, but they also acknowledged that the usability of these markers can be influenced by many factors such as the age of the twins, their gender, the tissue type et cetera. This may lead to the need for separation of markers based on the gender of the donor and their age range. What they also found is that using different methods can influence the number of markers found, which highlights the need for a standardized and highly validated protocol in forensic casework to ensure robustness. Lastly, it was found that many candidate markers are tissue-specific and therefore, the trace and reference samples collected in casework should be of the same tissue type to get the most reliable results. With blood or saliva this is feasible, but for other types of tissue encountered in crimes, one can imagine that this is more difficult to achieve.

4.2.3 Concluding remarks

So far, the epigenetic differences between monozygotic twins have been widely studied in many areas, but the forensic area is still lacking. Some studies have been published on the forensic feasibility of using several types of samples to study epigenetics, but there is still a general consensus that several more considerations need to be taken into account before application in forensic casework can become a reality. Validation of the methods is a crucial step, for which for example the reproducibility, efficiency and applicability of Bisulfite Sequencing in forensic casework need to be assessed. It is for example known that bisulfite conversion is not 100% efficient and this will only be complicated by the fact that crime samples are often of low quality or quantity. So in order for epigenetic analysis to be reliably implemented in forensic casework, the previously mentioned characteristics of the method need to be further explored. When applied in an accurate manner, one can imagine that epigenetics has the potential to become a very powerful method in forensics. This is mainly due to its high diversity, where in an ideal situation information on age, kinship and tissue type can be combined to achieve a very high value of evidence. Furthermore, marker discovery studies have already shown that differential epigenetic markers between twins are much more abundant than SNPs, allowing for higher discriminatory power. In the most ideal situation, SNPs and methylation patterns would even have the potential to be combined for an even higher value. Before this can all occur however, more studies need to be performed which specifically employ Bisulfite Sequencing and possibly other DNA methylation detection methods in a forensic environment to validate their use in casework. Additionally, it would be feasible to look into recurring differential CpG sites between monozygotic twins to aid the possible developments of standardized CpG kits.

14

5 DISCUSSION AND CONCLUDING REMARKS

It is estimated that in the Netherlands, around 1 in 250 pregnancies results in monozygotic twins (Nederlands Tweelingen Register, VU Amsterdam). Involvement of twins in a forensic case is not very common, but not unique either. Between 2002 and 2017, 577 twin DNA profiles were present in the Dutch DNA database, most of which contained profiles of both twins (Nederlands Forensisch Instituut, 2020). During this time period, there were 131 instances in which a twin profile matched a trace in a forensic investigation. Cases ranged from robberies to drug-related crimes and violence. In July 2020, the total database consists of 300.000 entries, of which ca. 700 are clusters of identical twins. Furthermore, there are two clusters present of identical triplets. When it comes to monozygotic twins, two extremes are often seen by the NFI: twins are either partners in crime or they point the finger at each other to cover for themselves. Furthermore, as they often look very much like one another, it is difficult for eye witnesses to distinguish between the two. To elucidate who is the true donor of the DNA in these cases when little other evidence is present, the current standard method of PCR-CE STR profiling is virtually useless, as STR profiles are identical between monozygotic twins. This fact has opened a whole new field of research into the possibility of distinguishing between monozygotic twins in a forensic context based on their DNA using other methods.

This thesis has discussed two of such possibilities. One of those is the use of SNPs: small point mutations that occur in the genomes of twins. Although a very promising method with the costs of Massively Parallel Sequencing approaches dropping over time, there are still many roadblocks that forensic researchers face when it comes to applying this in casework. Even though there is a high probability of SNPs arising between twins, there is no guarantee of them being present or being able to be detected. Furthermore, there is to this date no data on whether there is a correlation between SNPs of different twin pairs or whether they are random. The latter would complicate matters highly, as ultra-deep Whole Genome Sequencing would be necessary to detect these randomly occurring SNPs. This is a relatively costly method that requires processing of large amounts of data, would be difficult to standardize and brings along some privacy concerns. This is especially the case when DNA analyses not involving related people would suffice with a targeted sequencing method using a standardized kit, making the use of Whole Genome Sequencing a rare event in a forensic lab. However, when shared SNPs between pairs of twins can be found, this opens up the opportunity for development of standardized kits and more accessible analysis of twin DNA. On the other hand, the costs of ultra-deep Whole Genome Sequencing are dropping by the year, making the implementation of the method in forensics a more realistic option. However, until that happens, researchers will have to focus on SNP marker discovery and validation of Next Generation Sequencing methods for the forensic field to make genetic discrimination between monozygotic twins in forensic casework a reality.

The alternative approach discussed makes use of epigenetic patterns, a feature that has been known to be heavily dependent on external as well as internal factors and thus has a high probability of being significantly different between monozygotic twins. Several studies have been conducted to test the feasibility of this approach and they have lead to some promising CpG markers that would allow the development of forensic epigenetic analysis kits. However, it has also been acknowledged that epigenetic marks are highly problematic in the fact that they are often dynamic and specific to tissue type, age or gender. This may limit their usability in samples collected at different times or from different tissue types. To test this, more research into the specific forensic applicability of the discovered markers should be conducted. For example, their dynamics over time and their prevalence in several reference populations should be assessed. Thus far, there has been an overload of research into epigenetic markers

15 in the biomedical field in the context of diseases, but research in the forensic field is still lacking. For the latter, it is crucial that factors such as the efficiency and error rate of the method are assessed in a detailed manner, so that we can make a reliable conclusion on whether epigenetic markers for twin distinction may be used in forensic casework in the future.

One thing that we can generally conclude from this is that implementation of targeted Next Generation Sequencing methods in forensic casework in the near future is virtually inevitable. In fact, the first kit using such an approach has already been developed and thoroughly tested by independent laboratories. However, the question remains whether this approach will also be feasible for monozygotic twin cases. There is to this day no guarantee that the SNPs used by the ForenSeq kit, the SNPs discovered in the study by Weber-Lehmann et al. and the proposed CpG markers are widely recurring in other twin pairs. The solution to this problem depends heavily on novel marker discovery and screening, in the genetic field as well as the epigenetic field. If future research leads to discovery of widely present differential SNP or CpG markers between monozygotic twins and other related people, this would facilitate the use of a similar targeted NGS approach as that described above in twin cases. However, as long as this problem remains unsolved, we will have to rely on the more complicated Whole Genome Sequencing approach to find sparse differences in the DNA of twins. And although this will likely not become the standard practice for forensic laboratories due to the costs and the big, unstandardized data that come along with it, its high coverage may just give us the answers that we have been searching for.

In the words of Dr. Michael Krawczak: “It’s not something that’s going to happen every day in every laboratory. But once people become aware of this, there may be a lot of cold cases that come back to life.”

16

6 REFERENCES

Alberts, B., Johnson, A., Lewis, J., Morgan, D., Raff, M., & Keith Roberts, P. W. (2018). Molecular biology of the cell.

Baranzini, S. E., Mudge, J., Van Velkinburgh, J. C., Khankhanian, P., Khrebtukova, I., Miller, N. A., ... & May, G. D. (2010). Genome, epigenome and RNA sequences of monozygotic twins discordant for multiple sclerosis. Nature, 464(7293), 1351-1356.

Behjati, S., & Tarpey, P. S. (2013). What is next generation sequencing?. Archives of Disease in Childhood-Education and Practice, 98(6), 236-238.

Børsting, C., & Morling, N. (2015). Next generation sequencing and its applications in forensic genetics. Forensic Science International: Genetics, 18, 78-89.

Bruder, C. E., Piotrowski, A., Gijsbers, A. A., Andersson, R., Erickson, S., de Ståhl, T. D., ... & Crowley, M. (2008). Phenotypically concordant and discordant monozygotic twins display different DNA copy-number-variation profiles. The American Journal of Human

Genetics, 82(3), 763-771.

Butler, J. M. (2009). Fundamentals of forensic DNA typing. Academic press.

Butler, J. M. (2014). Advanced topics in forensic DNA typing: interpretation. Academic Press. Caskey, C. T., & Edwards, A. O. (1994). U.S. Patent No. 5,364,759. Washington, DC: U.S. Patent and Trademark Office.

Dal, G. M., Ergüner, B., Sağıroğlu, M. S., Yüksel, B., Onat, O. E., Alkan, C., & Özçelik, T. (2014). Early postzygotic mutations contribute to de novo variation in a healthy monozygotic twin pair. Journal of medical genetics, 51(7), 455-459.

Fraga, M. F., Ballestar, E., Paz, M. F., Ropero, S., Setien, F., Ballestar, M. L., ... & Boix-Chornet, M. (2005). Epigenetic differences arise during the lifetime of monozygotic twins. Proceedings of the National Academy of Sciences, 102(30), 10604-10609. Gjertson, D. W., Brenner, C. H., Baur, M. P., Carracedo, A., Guidet, F., Luque, J. A., … Morling, N. (2007). ISFG: Recommendations on biostatistics in paternity testing. Forensic Science International: Genetics, 1(3-4), 223–231.

Grunert, M., Appelt, S., Grossfeld, P., & Sperling, S. R. (2020). The Needle in the Haystack—Searching for Genetic and Epigenetic Differences in Monozygotic Twins

Discordant for Tetralogy of Fallot. Journal of Cardiovascular Development and Disease, 7(4), 55.

Jeffreys, A. J., Wilson, V., & Thein, S. L. (1985). Hypervariable ‘minisatellite’regions in human DNA. Nature, 314(6006), 67-73.

Jeffreys, A. J., Wilson, V., Neumann, R., & Keyte, J. (1988). Amplification of human minisatellites by the polymerase chain reaction: towards DNA fingerprinting of single cells. Nucleic Acids Research, 16(23), 10953-10971.

Köcher, S., Müller, P., Berger, B., Bodner, M., Parson, W., Roewer, L., ... & DNASeqEx Consortium. (2018). Inter-laboratory validation study of the ForenSeq™ DNA Signature Prep Kit. Forensic Science International: Genetics, 36, 77-85.

17 Krawczak, M., Cooper, D. N., Fändrich, F., Engel, W., & Schmidtke, J. (2012). How to

distinguish genetically between an alleged father and his monozygotic twin: a thought experiment. Forensic Science International: Genetics, 6(5), e129-e130.

Krawczak, M., Budowle, B., Weber-Lehmann, J., & Rolf, B. (2018). Distinguishing genetically between the germlines of male monozygotic twins. PLoS genetics, 14(12), e1007756.

Kunio, M., Yang, C., Minakuchi, Y., Ohori, K., Soutome, M., Hirasawa, T., ... & Goto, Y. I. (2013). Comparison of genomic and epigenomic expression in monozygotic twins discordant for Rett syndrome. PLoS One, 8(6), e66729.

Li, C., Zhang, S., Que, T., Li, L., & Zhao, S. (2011). Identical but not the same: the value of DNA methylation profiling in forensic discrimination within monozygotic twins. Forensic

Science International: Genetics Supplement Series, 3(1), e337-e338.

Liu, L., Li, Y., Li, S., Hu, N., He, Y., Pong, R., ... & Law, M. (2012). Comparison of next-generation sequencing systems. BioMed research international, 2012.

Metzker, M. L. (2010). Sequencing technologies—the next generation. Nature reviews

genetics, 11(1), 31-46.

Molenaar, D. (2020, August 3). Tweelingbroer verdachte verkrachting bejaarde vrouw uit Drenthe werkt toch mee aan extra DNA-onderzoek (en dat is uniek in Nederland). Dagblad van het Noorden. Retrieved from https://www.dvhn.nl/drenthe/Tweelingbroer-verdachte-verkrachting-bejaarde-vrouw Drenthe-werkt-toch-mee-aan-extra-DNA-onderzoek-25898790.html?harvest_referrer=https%3A%2F%2Fdnadatabank.forensischinstituut.nl%2

Motahari, A. S., Bresler, G., & David, N. C. (2013). Information theory of DNA shotgun sequencing. IEEE Transactions on Information Theory, 59(10), 6273-6289.

Mullis, K., Faloona, F., Scharf, S., Saiki, R. K., Horn, G. T., & Erlich, H. (1986, January). Specific enzymatic amplification of DNA in vitro: the polymerase chain reaction. In Cold Spring Harbor symposia on quantitative biology (Vol. 51, pp. 263-273). Cold Spring Harbor Laboratory Press.

Nederlands Forensisch Instituut (2020). Kansen om eeneiige tweelingen in strafzaken te

onderscheiden. @NFI. Retrieved from

http://magazines.forensischinstituut.nl/atnfi/2020/34/kansen-om-eeneiige-tweelingen-in-strafzaken-te-onderscheiden#:~:text=Eeneiige tweelingen hebben identieke standaard DNA-profielen.&text=In de periode 2002-2017,een misdrijf of veroordeeld worden.

Nederlands Tweelingen Register, VU Amsterdam. Retrieved from https://tweelingenregister.vu.nl/

Nishioka, M., Bundo, M., Ueda, J., Yoshikawa, A., Nishimura, F., Sasaki, T., ... & Iwamoto, K. (2018). Identification of somatic mutations in monozygotic twins discordant for psychiatric disorders. npj Schizophrenia, 4(1), 1-7.

Ogino, S., Kawasaki, T., Brahmandam, M., Cantor, M., Kirkner, G. J., Spiegelman, D., ... & Fuchs, C. S. (2006). Precision and performance characteristics of bisulfite conversion and real-time PCR (MethyLight) for quantitative DNA methylation analysis. The Journal of molecular

18 Park, J. L., Woo, K. M., Kim, S. Y., & Kim, Y. S. (2017). Potential forensic application of DNA methylation to identify individuals in a pair of monozygotic twins. Forensic Science International: Genetics Supplement Series, 6, e456-e457.

Rolf, B., & Krawczak, M. (2020). The germlines of male monozygotic (MZ) twins: Very similar, but not identical. Forensic Science International: Genetics, 102408.

Sanger, F., Nicklen, S., & Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proceedings of the national academy of sciences, 74(12), 5463-5467.Schuster, S. C. (2008). Next-generation sequencing transforms today's biology. Nature methods, 5(1), 16-18.

Sharma, V., van der Plaat, D. A., Liu, Y., & Wurmbach, E. (2020). Analyzing degraded DNA and challenging samples using the ForenSeq™ DNA Signature Prep kit. Science &

Justice, 60(3), 243-252.

Silvia, A. L., Shugarts, N., & Smith, J. (2017). A preliminary assessment of the ForenSeq™ FGx System: next generation sequencing of an STR and SNP multiplex. International journal

of legal medicine, 131(1), 73-86.

State v. Miller 91 S.W.3d 630 (Missouri Court of Appeals, 2007)

Vidaki, A., & Kayser, M. (2018). Recent progress, methods and perspectives in forensic epigenetics. Forensic Science International: Genetics, 37, 180-195.

Voelkerding, K. V., Dames, S. A., & Durtschi, J. D. (2009). Next-generation sequencing: from basic research to diagnostics. Clinical chemistry, 55(4), 641-658.

Vogt, J., Kohlhase, J., Morlot, S., Kluwe, L., Mautner, V. F., Cooper, D. N., & Kehrer‐Sawatzki, H. (2011). Monozygotic twins discordant for neurofibromatosis type 1 due to a postzygotic NF1 gene mutation. Human mutation, 32(6), E2134-E2147.

Weber-Lehmann, J., Schilling, E., Gradl, G., Richter, D. C., Wiehler, J., & Rolf, B. (2014). Finding the needle in the haystack: differentiating “identical” twins in paternity testing and forensics by ultra-deep next generation sequencing. Forensic Science International: Genetics, 9, 42-46.

Westen, A. A., Kraaijenbrink, T., de Medina, E. A. R., Harteveld, J., Willemse, P., Zuniga, S. B., ... & Sijen, T. (2014). Comparing six commercial autosomal STR kits in a large Dutch population sample. Forensic Science International: Genetics, 10, 55-63.

Zimmer, C. (2019, March 1). One Twin Committed the Crime – but Which One? A New DNA Test Can Finger the Culprit. The New York Times. Retrieved from https://www.nytimes.com/2019/03/01/science/twins-dna-crime-paternity.html

19

7 APPENDIX: SEARCH STRATEGY

To provide context information in the introduction and discussion of this thesis, several newspaper and magazine articles were retrieved using Google and the website of the Netherlands Forensic Institute as the main search engines. One magazine article and two newspaper articles were used for this thesis. Furthermore, the Dutch Twin Database provided by the Vrije Universiteit Amsterdam was used.

❖ Time span of searches: October 26th _{2020 – November 30}th ₂₀₂₀

❖ Language restriction: Newspaper and magazine articles used for this thesis are written in both Dutch and English.

❖ Publication year restrictions: Attention was given to providing the most recent reference cases and statistics for context. The publication years of the articles used are 2019 and 2020.

Search terms used:

Dutch

‘Tweeling statistieken’

‘NFI’ ‘Nederlands Forensisch Instituut’ ‘NFI tweeling’ ‘NFI tweelingen’

‘Drenthse zedenzaak’ ‘Drenthe tweelingen zaak’ ‘Eeneiige tweelingen’ ‘Nederland eeneiige tweelingen’

English

‘Forensic twins’

‘Forensic DNA analysis twins’

‘McNair twins’ ‘McNair David Deakin’ ‘Mcnair brothers case’ ‘Hollie Marie Adams paternity’ ‘Missouri court of appeals adams’

To search for scholarly articles, books and court decisions, Google Scholar was used as a main database. For this thesis, a total of 3 books, 31 scholarly articles and 1 court decision were used.

❖ Time span of searches: October 26th _{2020 – December 14}th ₂₀₂₀

❖ Language restrictions: All scholarly articles and books used for this thesis are written in English.

❖ Publication year restrictions: No restrictions were put on the year of publication for scholarly articles and books, although attention was given to reporting the most recent and relevant results on the topics in question. The publication years used range from 1985 – 2020.

❖ Peer-review restrictions: All scholarly articles used in this thesis have been published in peer-reviewed journals.

20 The following peer-reviewed journals have been used:

o Nature (including Nature reviews Genetics and Nature Methods) o Archives of Disease in Childhood-Education and Practice

o Forensic Science International: Genetics o American Journal of Human Genetics

o Journal of Cardiovascular Development and Disease o Proceedings of the National Academy of Sciences (PNAS) o Nucleic Acids Research

o PLoS One and PLoS Genetics o BioMeds Research International

o IEEE Transactions on Information Theory o Cold Spring Harbor Laboratory Press o NPJ Schizophrenia

o The Journal of Molecular Diagnostics o Science & Justice

o International Journal of Legal Medicine o Clinical Chemistry

o Human Mutation

Below is an overview of the most important search terms used:

Forensic DNA analysis

‘STR Analysis’ ‘Alec Jeffreys’ ‘Polymerase Chain Reaction’

‘Fundamentals of forensic DNA typing’ ‘STR kits’

‘Restriction fragment length polymorphism’ ‘Short tandem repeats’

Epigenetics

‘Epigenetics techniques’ ‘Bisulfite conversion’ ‘bisulfite’

‘Forensic methylation’ ‘Forensic epigenetics’ ‘Forensic epigenetics twins’

‘Epigenetics twins’

‘Molecular biology of the cell’

Next Generation Sequencing and Genetics

‘Dideoxy Sanger’ ‘Dideoxy Sequencing’ ‘Forensic twins’ ‘Twin SNPs’ ‘Twin STRs’ ‘ForenSeq’

‘Next Generation Sequencing’ ‘Illumina HiSeq’

‘Krawczak thought experiment’ Michael Krawczak 2012’ ‘NGS forensics’