CONTRAST: a discriminative, phylogeny-free approach to multiple informant de novo gene prediction

Samuel S Gross¹ , Chuong B Do¹ , Marina Sirota² and Serafim Batzoglou¹

¹Computer Science Department, Stanford University, Stanford, CA, USA

²Biomedical Informatics, Stanford University, Stanford, CA, USA

author email corresponding author email

Genome Biology 2007, 8:R269doi:10.1186/gb-2007-8-12-r269


Published:	20 December 2007

Subject areas: Bioinformatics, Genetics, Genome studies

Abstract

We describe CONTRAST, a gene predictor which directly incorporates information from multiple alignments rather than employing phylogenetic models. This is accomplished through the use of discriminative machine learning techniques, including a novel training algorithm. We use a two-stage approach, in which a set of binary classifiers designed to recognize coding region boundaries is combined with a global model of gene structure. CONTRAST predicts exact coding region structures for 65% more human genes than the previous state-of-the-art method, misses 46% fewer exons and displays comparable gains in specificity.