Cutting-edge computational molecular biology research featured in Genome Research

Genome Research is publishing several papers in coordination with the Research in Computational Molecular Biology (RECOMB) 2008 Conference, March 30, 2008 – April 2, 2008, at the National University of Singapore. Genome Research has partnered with RECOMB to publish a select number of high-quality contributions to the meeting, presenting the latest theoretical advances in computational biology and their applications in molecular biology and medicine. The papers will appear online Wednesday, March 19, 2008, and in print as a special section of the Genome Research April 2008 issue.

1. Inferring ancestral genomes from admixed populations

Understanding the origins, migrations, adaptations, and admixtures of human populations is often severely complicated by a lack of documentation or archaeological evidence. However, the genomic structure of an individual or a population can serve as a biological record of ancestry. Endeavors such as the International HapMap Project have made considerable progress in describing common patterns of human genetic variation, yet analyzing this data to inform upon ancestry remains a formidable task. Two papers published in Genome Research present new mathematical models directly addressing this challenge.

Led by Dr. Serafim Batzoglou of Stanford University, researchers have designed a model that significantly improves scientists’ ability to determine ancestry based upon genomic features. “We created a new model that improved accuracy to such an extent that distinguishing between the continental populations in HapMap became possible up to 20 generations back,” describes co-first author Andreas Sundquist. Though genetic variation data from closely related populations is lacking, Sundquist and co-first author Eugene Fratkin overcame this obstacle by constructing simulated population sets to test their model. “We were then able to conduct tests on these populations and analyze the accuracy of our method as a function of both the population divergence and the number of generations of admixture,” explains Sundquist. “Our results show that challenges still remain in distinguishing between closely related populations, but that we have vastly improved the state-of-the-art.”

In the second paper, researchers from the International Computer Science Institute and the University of California, Berkeley, have taken another approach to inferring ancestry. Sankararaman et al. present a new computational model that improves on previous methods for deriving ancestry information from admixed populations, by more accurately modeling linkage disequilibrium and predicting historical recombination events. The authors utilize their algorithm to tackle problems such as inferring locus-specific ancestry in a population derived from unknown ancestral populations.