Category Archives: Forensics

When DNA Isn’t Enough: Methylation, Forensics, and Twins

DNA evidence is often considered a “home run” in forensics. If you find readable DNA at the crime scene, and it matches a suspect, a correct conviction is almost assured. A DNA sample can often point to a single individual with ridiculous specificity – often 1 in a quadrillion or greater. But, what happens when someone else shares your DNA?

Monozygotic, or “identical” twins differ from dizygotic, or “fraternal” twins in that they come from the same zygote, hence, “mono”zygotic. In other words, identical twins come from 1 fertilized egg, while fraternal come from two. This means that Identical twins will share the same DNA, while fraternal twins will share as much DNA as any other sibling pair. There are, of course, many iterations of monozygosity depending when during development the split actually takes place. This nuance has led scientists in Germany to a possible solution to the issue of identical twin DNA.

During development, only a few cells are present. These cells begin to differentiate into the different tissue types that they will become. As these cells divide rapidly to produce the all of the daughter cells, mutations can occur in the DNA. If the mutation occurs earlier, it will be present in a larger ratio of the daughter cells, and will be more easily detectable during the twin’s lifetime. This differentiation of tissues also means that, the earlier the twins split, the less mutations they will have in common (and, thus, the more differences you can detect in their DNA). It has been suggested recently that, a handful of single nucleotide mutations, or “SNPs” can be found between twins. However, these SNPs aren’t so easy to find in a sea of 3 billion other nucleotides. To find these few differences and find them reliably, the entire genome of both twins must be sequenced several times over. In the case of the German scientists, their experiment results in 94-fold coverage, meaning they covered each of the 3 billion nucleotides 94 times. This must be done to ensure accuracy. At 3 billion nucleotides, a 99.9% accuracy will still result in 30 million errors. If anything, this shows how incredibly accurate our cellular machinery is.

At any rate, the scientists tested their new method on a set of twins, and it worked. In the end, twelve SNPs were identified between the twin brothers. Typically, one experiment is not considered to hold much weight in science, but this particular experiment is backed by strongly reinforced genetic theory, and the results were exactly what we would expect.

So, case solved, right? Well, maybe not. It turns out that this method comes with a hefty price tag – over $100,000. This is far too much to be practical in forensic case work, especially when you consider that about 1 person in 167 is an identical twin. Of course, this price will go down as DNA sequencing prices continue to plummet in light of newer, better technology. Still, it will be many years before anything like this will be affordable (a typical forensic DNA test costs in the neighborhood of $400-$1000). Furthermore, the instruments used in this method (Next generation sequencing), though typical in research science, have not been approved for use in court. That in and of itself can be a challenging obstacle to overcome, regardless of costs.

Perhaps in a few decades these issues will be resolved. Perhaps not. Either way, it might be a good idea to have a plan in the meantime. This is (hopefully) where my master’s thesis comes in.

DNA is composed of four nucleotides, commonly noted as “A T C and G.” Throughout life, a methyl group – a carbon and three hydrogens – attaches to some of the C’s in your genome. This is known as DNA methylation, which is a big component of the larger phenomenon known as epigenetics. As it turns out, these methyl groups attach randomly to the C’s, though some evidence suggests that environmental conditions may play some part in this. In any case, the attaching of methyl groups to C’s is different among individuals – even identical twins. In fact, studies have shown that newborns already exhibit DNA methylation discordance. Presumably, these differences would become more pronounced as time goes on. Not many studies have looked at this, but the ones that have also show evidence of greater discordance with age.

There is a potential issue with studying DNA methylation: it doesn’t occur uniformly among tissues. In other words, a blood sample and a skin sample from the same individual will show different patterns of methylation. Moreover, cells within the same tissue can show different methylation patterns. Though not insurmountable, these issues make methylation analysis a tricky subject.

To tackle the first issue of tissue discordance, one could simply match the type of DNA you take from the suspect with the type of DNA you have at the scene. The second issue of intra-tissue discordance is a bit trickier to tackle. For starters, we don’t know too terribly much about how DNA methylation works. Ostensibly, if methylation differences occurred early in development, then they would show the same pattern of proliferation as the SNPs that occur early in development. This means that the same DNA methylation pattern would be present in all of the daughter cells, and show up easily in a DNA sample from that tissue.

Another possible solution would be to take a statistical approach. This would involve looking at the methylation patterns several times and coming up with an “average” methylation. For example, let’s say there are 10 C’s susceptible to methylation in a particular DNA sequence. If I run 10 samples from a DNA swab, I might find the number of methylated C’s to be: 3, 4, 5 ,3, 2, 4, 5, 3, 4, 4. If you average these, you get 3.7 out of 10 possible methylated C’s. Thus, you might say that this DNA sequence shows 37% methylation. If you do the same thing for the other twin and come up with 5.5 out of 10 possible methylated C’s, you could say that the other twin’s sequence shows 55% methylation. Ideally, these number would be relatively reproducible, especially as you increase the sample number and/or number of potentially methylated C’s per sequence.

Compared to the SNP method, my project is less definitive. However, good protocols would still make the method definitive enough. Once you narrow the suspects down to two twins via normal DNA testing, you have two possible outcomes: a match between one twin and the sample at the crime scene, or inconclusive. At this point, you just need to differentiate between two people, not 7 billion. Thus, the required statistical power is much, much lower. The big difference between my method and the SNP method is the price. Whereas the SNP method costs between $100,000 and $160,000, my method could be done in-house for less than $5000. Furthermore, my method is performed using the same instruments as traditional DNA testing, meaning that the new instrumentation does not need to be validated for use in court.

So, while it will take some work, and my project is more of a proof of concept study, the use of DNA methylation in forensics is generating a lot of attention. One of the issues with methylation in my study, i.e., different patterns in different tissues, has been a major benefit to a different use of DNA methylation – tissue identification. The idea here being that if you can identify consistent methylation patterns among a tissue type, you can use those patterns to identify the tissue. Another aspect that is relevant to my project, the increase in methylation with age, has been vetted as a possible investigative tool. If you can identify level of methylation that are consistent with different age groups, you can potentially “age” a suspect just by their DNA methylation. Studies on methylation aging are few and far between, but preliminary results are promising, suggesting that age-based methylation analysis can get within +/- 5 years of an individual’s actual age.

As we learn more about DNA methylation, it will become more useful. This is true not only for forensics, but also medicine, since methylation plays an important role in turning genes “on” or “off.” This is particularly true in cancer, where abnormal DNA methylation seems to occur. But, before we try to cure cancer with methylation, perhaps we can perform the smaller task of telling two twins apart from each other.

*Also published in part at http://forensicoutreach.com/library/when-dna-isnt-enough-methylation-forensics-and-twins-part-1/

and

http://forensicoutreach.com/library/when-dna-isnt-enough-methylation-forensics-and-twins-part-2/

Advertisements