Lipid: a linear programming approach for isoform detection and abundance estimation

Alternative splicing of the premature mRNA is an important step of gene expression regulation a ecting 75%-95% of human genes. Multiple studies have clearly demonstrated that compared to normal tissues, tumours shift splicing patterns of many cancer associated genes, which involves either complete switch from one isoform to another or change in ratio of isoforms. During the past decade many algorithms have been proposed to detect splice isoforms using high resolution microarrays and sequencing data. However, inferring relative abundance of detected isoforms remains a challenge. Here we present a Linear Programming method that infers splice isoforms expressed in a given sample and estimates their absolute abundance. The algorithm is applicable to any sub-gene level expression data from both micro-array and RNA-Seq technologies and requires exon annotation. We aim the optimization function at minimizing the deviation from the expression of the regions, and maximizing the lengths of the isoforms.