VSEARCH

VSEARCH performs high-throughput nucleotide sequence processing and analysis for metagenomics, genomics, and population genomics, including searching, global alignment, clustering, dereplication, chimera detection, and FASTQ processing.

Key Features:

• Multithreading and performance: 64-bit multithreaded application with vectorisation to leverage parallel processing for large datasets and alignment tasks.
• Heuristic-based searching: Fast heuristic search based on shared words between query and target sequences.
• Global sequence alignment: Full dynamic programming for optimal global sequence alignments.
• Clustering and dereplication: Clustering by similarity with length pre-sorting, abundance pre-sorting, or user-defined order, and dereplication of full-length sequences or prefixes.
• Chimera detection: Both reference-based and de novo chimera detection methods.
• Pairwise alignment: Parallel computation of pairwise alignments.
• FASTQ processing: FASTQ format detection, filtering, read quality statistics, merging of paired reads, shuffling, rereplication, and conversion between FASTQ formats.
• Low-complexity masking: Masking of low-complexity sequences using the DUST algorithm.
• Sequence operations: Reverse complementation, sorting, and subsampling of sequences.
• Similarity definitions: Choice among various similarity definitions to tailor analyses.
• Comparative performance: Demonstrates superior accuracy in searching, clustering, chimera detection, and subsampling compared to USEARCH, while being slower than USEARCH for clustering and chimera detection but faster for paired-end read merging and dereplication.

Scientific Applications:

• Metagenomics: Accurate sequence processing, chimera detection, clustering, and paired-read merging for metagenomic studies.
• Genomics: Nucleotide sequence alignment, clustering, and dereplication for genomics analyses.
• Population genomics: Dereplication and subsampling to support population genomics workflows.
• Paired-end read processing: Merging and downstream processing of paired reads, including quality filtering and format conversion.
• Chimera detection and clustering: Reference-based and de novo chimera detection combined with similarity-based clustering for sequence-based analyses.

Methodology:

Methods explicitly include heuristic searching based on shared words, full dynamic programming for global alignment, clustering by similarity with length/abundance/user-defined sorting, full-length and prefix dereplication, reference-based and de novo chimera detection, parallel pairwise alignments, multithreading and vectorisation, FASTQ format detection and filtering, read quality statistics, paired-read merging, shuffling, rereplication, DUST low-complexity masking, reverse complementation, subsampling, and FASTQ format conversion.