HMMSplicer: A Tool for Efficient and Sensitive Discovery of Known and Novel Splice Junctions in RNA-Seq Data

• Published in: PloS one Vol. 5; no. 11; p. e13875
• Main Authors: Dimon, Michelle T., Sorber, Katherine, DeRisi, Joseph L.
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 08.11.2010; Public Library of Science (PLoS)
• Subjects: Acids; Algorithms; Alternative splicing; Animals; Arabidopsis - genetics; Arabidopsis thaliana; Biochemistry; Biophysics; Computational Biology - methods; Computational Biology/Alternative Splicing; Computational Biology/Genomics; Datasets; Erythrocytes; Gene expression; Gene Expression Profiling; Gene mapping; Gene sequencing; Genes; Genomes; Genomics; Humans; Isoforms; Markov analysis; Medical research; Molecular Biology/Bioinformatics; Molecular Biology/RNA Splicing; Next-generation sequencing; Organisms; Plasmodium falciparum; Plasmodium falciparum - genetics; Proteins; Reproducibility of Results; Ribonucleic acid; RNA; RNA sequencing; RNA Splice Sites - genetics; RNA Splicing - genetics; Splice junctions; Statistical analysis; Studies; United States > US; San Francisco California; California
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0013875

High-throughput sequencing of an organism's transcriptome, or RNA-Seq, is a valuable and versatile new strategy for capturing snapshots of gene expression. However, transcriptome sequencing creates a new class of alignment problem: mapping short reads that span exon-exon junctions back to the reference genome, especially in the case where a splice junction is previously unknown. Here we introduce HMMSplicer, an accurate and efficient algorithm for discovering canonical and non-canonical splice junctions in short read datasets. HMMSplicer identifies more splice junctions than currently available algorithms when tested on publicly available A. thaliana, P. falciparum, and H. sapiens datasets without a reduction in specificity. HMMSplicer was found to perform especially well in compact genomes and on genes with low expression levels, alternative splice isoforms, or non-canonical splice junctions. Because HHMSplicer does not rely on pre-built gene models, the products of inexact splicing are also detected. For H. sapiens, we find 3.6% of 3' splice sites and 1.4% of 5' splice sites are inexact, typically differing by 3 bases in either direction. In addition, HMMSplicer provides a score for every predicted junction allowing the user to set a threshold to tune false positive rates depending on the needs of the experiment. HMMSplicer is implemented in Python. Code and documentation are freely available at http://derisilab.ucsf.edu/software/hmmsplicer.