WTFgenes: What's The Function of these genes? Static sites for model-based gene set analysis

A common technique for interpreting experimentally-identified lists of genes is to look for enrichment of genes associated to particular ontology terms. The most common technique uses the hypergeometric distribution; more recently, a model-based approach was proposed. These approaches must typically be run using downloaded software, or on a server. We develop a collapsed likelihood for model-based gene set analysis and present WTFgenes, an implementation of both hypergeometric and model-based approaches, that can be published as a static site with computation run in JavaScript on the user's web browser client. Apart from hosting files, zero server resources are required: the site can (for example) be served directly from Amazon S3 or GitHub Pages. A C++11 implementation yielding identical results runs roughly twice as fast as the JavaScript version. WTFgenes is available from https://github.com/evoldoers/wtfgenes under the BSD3 license. A demonstration for the Gene Ontology is usable at https://evoldoers.github.io/wtfgo . Contact: Ian Holmes [email protected] .