 Imagine that you're the victim of a crime. There's biological evidence but investigators are not able to identify the donor of that DNA sample and the case goes cold. This is the reality for many cases in New Zealand and worldwide and this is something that my research hopes to change. Good evening, I'm Rebecca. I'm a Forensic Science PhD student at the University of Auckland but based at ESR which is the Crown Research Institute in New Zealand that provides forensic services to police. And my research looks at how we can use DNA methylation or CPG markers and forensic applications namely to help identify the donor of a DNA sample. So just to begin, what is DNA methylation? So DNA is made up of four bases, adenine, thymine, cytosine and guanine and DNA methylation is simply the addition of a methyl group to these bases. It most almost exclusively occurs at the cytosine base in humans to form five-metallic cytosine and it most commonly occurs in the context of a CPG site so that's when a cytosine is followed by a guanine. It's an epigenetic modification which like you've heard influences gene expression. It's also stable and it's preserved during cellular division which makes it a good biomarker. But what's really cool about DNA methylation is that it's influenced by a wide variety of environmental and lifestyle factors. This includes exercise, medication, diet, tobacco and alcohol consumption as well as stress levels in age. So why is this of interest in forensics? Well currently if a crime occurs and biological evidence is left at a scene we can test this to get the DNA profile and then we can compare that crime scene DNA profile to DNA profiles of known people. If a match occurs we can link that person to our crime scene evidence. However a problem arises when that crime scene DNA profile actually links to two different people and this is what happens in the case of identical twins. Because identical twins have come from the same fertilised egg and have the same DNA profile which means that we cannot tell which twin deposited a DNA sample and that obviously poses problems from investigators and in court. However because DNA methylation is known to be influenced by those wide range of factors although they might have the same DNA profile identical twins should show differences in their DNA methylation patterns and using this we could differentiate them. The other issue that occurs in forensic biology is when our crime scene sample doesn't match to any known person. Now in these cases we usually can't give investigators much more information. However one of the things that DNA methylation is known to be influenced by is age. So if we could find DNA methylation correlated markers that are correlated with age we could be able to perform a prediction model and actually estimate the age of a person from their DNA and provide this intelligence information to investigators. So these are the two applications that I'm interested in micro-research. Literature has shown that it is possible and we wanted to establish could we do this in the New Zealand population and could we do it using a methodology that could be easily implemented into our forensic lab. So the objectives of my research are first to identify CPG or methylation markers that are either correlated with age or could be used to differentiate identical twins. Develop these into marker panels and use them in forensically relevant body fluids. Test in a New Zealand population and then we also want to explore the variables that could impact the accuracy and finally validate this methodology for case work. So I began by collecting samples from either individuals of a wide range of ages to identify identical twins. I extracted the DNA. For my identical twins we underwent standard DNA profiling to in fact establish that they were identical. And then I underwent a process for DNA methylation profiling which involves a lot of fancy instruments but ultimately it involves these three steps and what's important is that these three steps in combination has been used in clinical epigenetics but at the time of this research they weren't really used in forensic research. However, this is the process we wanted to use because it was the best integrated into our forensic lab. The output of that is sequencing data which I then had to bioinformatically process to determine the level of methylation for each of my markers and then I did some statistical analysis. So for age estimation that was using correlation and regression analysis and for identical twin differentiation that was using t-test and p-values. So my proof of concept study had five pairs of identical twins and 28 individuals of a range of ages. My individuals were roughly evenly distributed across my different age groups. I collected buckle which are cheek sale samples, blood finger prick samples and semen where possible from my participants. And then I developed four different multiplexes. So multiplexes are a way that we can look at a whole bunch of markers from the same DNA sample and that's really important in forensic analysis because we often don't have a lot of DNA to work with. So two of these were focused on age correlated markers in blood. One was focused on age correlated markers in semen and one was focused on my twin differentiation markers. So now for some results I ended up looking at 84 different CPG markers for my 28 individuals from my blood samples and some of these, as you can see, showed quite a decent correlation with age. I then used a process called step-wise regression to identify the best combinations of markers to put in my prediction model. I ended up developing this linear regression model which had three markers and that had an adjusted R-squared of 0.87. So R-squared is a way we can measure how well the model is working with one being completely perfect and higher being better. This model also had a mean average deviation of 3.262 years and what that means is that on average my predicted age was just over three years away from my observed age of my participants. This research was very comparable to what had previously been reported in the literature from other studies that used other methodology. So that was promising for our methodology. I did the same thing with my semen samples. However, in this case, I only had samples from eight individuals. I looked at 10 sites. Again, some of them showed a really nice correlation with age. I ended up developing a linear regression model that had one site in it and that had a mean average deviation of just over four years. And again, that was pretty comparable with what had been reported in the literature. So finally, I also wanted to see if I could differentiate my identical twins. I used buckle samples. I looked at 31 different sites and what I found was that there was at least one site in all of my twin pairs that showed significantly different levels of methylation at the 5% significance level. However, there was no site that was consistently different across all of my twin pairs. Now, if I just draw your attention to this graph, what was really interesting was if you can look at the second site in, you see that the level of methylation was quite different between twin A and twin B. Whereas in the last site, it was actually really, really small, despite still being statistically significant. So this brought up a lot of questions for us regarding if we were going to take this forward into casework, do we need to start thinking about some thresholds for the minimum number of differences or magnitude of those differences. So in conclusion, I would love to say yes, we can use these methylation markers to help with identification, but in reality, we're still at the early stages. So I'm going to limit it by just saying it shows promise at this stage in the New Zealand population. There's still a lot of work to be done. So some of the things that I'm looking at is developing my methodology and my statistical models in particular, particularly maybe looking at some machine learning algorithms. I obviously want to test in a much, much larger sample size with a wider age range including children. I'm also really interested in investigating how those methylation patterns might change over time. And this is going to be really important say if a crime scene sample is left in January, you find your person of interest in July, they turn out to be an identical twin, how much variation are we expecting to see between those samples if that twin is the true donor. And finally, I want to perform a whole bunch of validation studies so we could actually use this in court not only in New Zealand, but also worldwide. I want to thank my knowledgements for my funding, my support, my supervisors and also my volunteers. Thank you.