Two 23andMe scientists have improved a long-used method for detecting shared DNA segments that researchers use to study everything from whether two people are related, to human demography, and even the heritability of disease.
The two computational biologists – Eric Durand, PhD and Cory McLean, PhD – recently published their findings in the journal Molecular Biology and Evolution. The pair made their algorithm, called HaploScore, publicly available.
Previous studies have used computer simulations to look at the accuracy of established methods for determining identity-by-descent (IBD), but Eric and Cory are the first to test those methods with real human data.
When they scrutinized current methods they found that the algorithms produced many errors, so Eric and Cory created a better and more accurate method for determining IBD. Their now open-source method is called HaploScore. “(HaploScore will) allow all researchers to more accurately identify genetic relationships between distantly-related individuals and allow for improved ancestry reports within 23andMe,” said Cory.
Methods for determining IBD are used by researchers to identify shared segments of DNA. Researchers look at IBD in many aspects of genetic research including determining the “relatedness” of any two individuals – the longer the shared segment(s), the closer the relationship. Up to now studies of the accuracy of the methods for determining IBD had been done with simulated data.
When Eric and Cory used the popular IBD detection method called GERMLINE on real data – almost 3,000 biological parent and child “trios” in the 23andMe database – they found it produced a false positive rate of about 67 percent for small segments of DNA. HaploScore improves the accuracy of short IBD segment detection, offering researchers a chance to look at much shorter segments of shared DNA to determine relatedness. Segments of DNA are measured in units called centiMorgans.
Eric and Cory’s new method allows for determining IBD more accurately on segments in the range of 2 to 4 centiMorgans. As more and more people are genotyped or have their DNA sequenced, having an accurate method for determining IBD will be increasingly important.