The Subread aligner: fast, accurate and scalable read mapping by seed-and-vote
Authors
Discussion
Loading...
Member takes
Implication
Computational study demonstrates fast and accurate genomic read mapping using a seed-and-vote strategy, indicating improved scalability for sequencing analysis.
Key Points
- To develop a fast, accurate, and scalable multi-seed strategy for mapping high-throughput sequencing reads to a reference genome.
- Engineered the 'seed-and-vote' algorithm (Subread), which extracts multiple short subreads from each sequencing read to vote on the optimal genomic location.
- Applied overlapping subreads for reads shorter than 160 base pairs and used conventional local alignment to resolve mismatches and indels within the winning consensus region.
- Extended the voting framework to detect split reads spanning exon junctions across distinct exons of the same gene.
- Determined candidate genomic locations directly from voting subreads prior to performing detailed base alignment, substantially increasing mapping speed.
- Achieved high mapping sensitivity and accuracy by requiring multi-subread consensus without constraining individual subreads to align contiguously.
- Demonstrated efficient scalability to longer sequencing read lengths and robust identification of complex exon junctions.
Cite This Study
Liao et al. (2013) studied this question.