Gene-finding machines
Every gene in the genome's autism association, ranked
As traditional sequencing-based studies have only identified a small fraction of the autism-linked genes predicted to exist, researchers from Princeton University and the Simons Foundation are turning to machine learning methods instead. In a study published in Nature Neuroscience last month, the team developed computational algorithms that cut through thousands of data sets to build a brain-specific genetic network and predict the autism relevance of every gene in the human genome.
"It's like a machine learning meta-analysis," said Olga Troyanskaya, one of the principal investigators behind the study. Troyanskaya is a professor of computer science at Princeton University, a member of the university's Lewis-Sigler Institute of Integrative Genomics and deputy director for genomics at the Simons Foundation...