In Science this week, a multinational team of researchers presented the results of an analysis of copy number variations across various human populations, offering insights into the forces that eliminate and maintain them. By studying CNVs across 236 individual genomes from 125 distinct human populations, the investigators found that DNA duplications have "fundamentally different population genetic and selective signatures" versus deletions, with the latter more reflective of selection and the former more likely to be stratified between populations. They were also able to identify patterns of ancestry and the maintenance of admixed genomes resulting from interbreeding between ancient hominids. GenomeWeb has more on this study here.
Also in Science, a group of researchers from Microsoft, IBM, Google, and elsewhere presents a new technique for preserving validity in adaptive data analysis even when reusing data. Based on insights from privacy-preserving data analysis, the method involves testing hypotheses against aggregate information while keeping certain information about specific dataset components confidential.