TransCoord: Genome-guided Transcripts Assembly by Coordinating Candidate Paths into Two-phased Linear Programming

High-throughput sequencing of mRNA (RNA-seq) provides a promise for transcriptome reconstruction by producing hundreds of millions of short reads. Current salient methods for genome-based transcriptome reconstruction almost unanimously sank into details instead of considering it in a whole picture. We present TransCoord, which inclusively gathers all kinds of candidate transcripts into a two-phased linear programming model that aims both to minimize coverage deviation and reserve least transcripts under conserving the must. In this way, the outcome is a coordination of all candidates, instead of a union of all independently assembled parts. Test on 19 human and 5 Arabidopsis thaliana real RNA-seq datasets, TransCoord outperformed all 4 compared salient assemblers in sensitivity. TransCoord is available at https://github.com/lcc121/TransCoord