Rsubread

Rsubread maps sequencing reads to reference genomes using a multi-seed "seed-and-vote" strategy to enable efficient, sensitive, and accurate alignment for genomic and transcriptomic analysis.

Key Features:

• Seed-and-vote strategy: Extracts multiple short seeds (subreads) from each read and uses their votes to determine the overall genomic location for the read.
• Subread extraction: Derives multiple subreads from each sequencing read to provide redundant mapping evidence.
• Preselection for efficiency: Selects the candidate genomic region before detailed alignment to speed up mapping.
• Mismatch and indel refinement: Applies conventional alignment algorithms to refine mismatches and indels within the preselected region.
• Sensitivity and flexibility: Does not require individual subreads to map exactly or be in close proximity, allowing accommodation of mapping variation.
• Consensus-based accuracy: Requires support from multiple subreads for a final mapped location, improving mapping accuracy.
• Exon junction detection: Identifies exon-exon junctions by detecting reads whose subreads map to different exons within the same gene.
• Scalability for long reads: Maintains performance and accuracy as read length increases.

Scientific Applications:

• Quality assessment: Assesses sequence read quality to inform downstream analysis.
• Read summarization: Summarizes mapped read data for interpretation of sequencing results.
• Indel detection: Detects short and long insertions and deletions during alignment refinement.
• Expression calling: Supports absolute expression quantification for transcriptomic studies.
• SNP calling: Enables detection of single nucleotide polymorphisms from aligned reads.

Methodology:

Performs a two-step alignment: (1) uses the seed-and-vote strategy where subreads vote to determine the genomic location of reads, and (2) applies conventional alignment algorithms to refine mismatches and indels within the selected region.