New gene expression algorithm developed

June 24, 2013

Researchers in the Institute for Brain & Neural Systems and have produced an algorithm for comparing two gene expression studies. The algorithm, built to address the complex problem of meta-analysis, is based on ranking the genes and computing the statistical likelihood of overlaps between sections of the ranked lists, and allows for the number of significant genes in the two experiments to be different (a 2-dimensional solution). It also removes the need for arbitrary thresholds in the test statistics when comparing experiments. In the a paper published in the Journal of Computational Biology, authors Michael Antosh, David Fox, Leon N Cooper and Nicola Neretti apply the algorithm to simulated and biological data sets, comparing the results with existing ranked list-based comparison algorithms.