CAT and BAT

CAT and BAT classify long DNA sequences and metagenome-assembled genomes (MAGs/bins) by assigning taxonomic labels using ORF-level alignments to the NCBI non-redundant (nr) protein database and a voting-based classification algorithm.

Key Features:

• Sequence scope: Works on individual long contigs and metagenome-assembled genomes (MAGs)/bins.
• Gene calling and ORF prediction: Performs gene calling and predicts open reading frames (ORFs) within input sequences.
• Protein alignments to nr: Maps predicted ORFs against the non-redundant (nr) protein database to obtain taxonomic signals.
• Voting-based classification: Aggregates ORF-level assignments using a voting-based algorithm to derive contig- or MAG-level taxonomy.
• Multiple-signal integration: Integrates multiple ORF signals instead of relying on single best-hit matches.
• Adaptive taxonomic rank assignment: Automatically assigns low-rank classifications when close relatives are present and higher-rank classifications when they are absent.
• Robustness to novel lineages: Provides precise taxonomy for sequences from highly unknown or deep-branching microbial lineages.

Scientific Applications:

• Taxonomic classification of contigs: Assign taxonomy to assembled long contigs from metagenomic datasets.
• Annotation of MAGs/bins: Provide taxonomic labels for metagenome-assembled genomes and genomic bins.
• De novo metagenomics: Support taxonomic profiling in de novo analyses of complex microbial communities.
• Novel lineage characterization: Facilitate placement and classification of novel or poorly characterized microbial taxa.

Methodology:

Performs gene calling and ORF prediction, maps predicted ORFs to the NCBI non-redundant (nr) protein database, and aggregates ORF-level hits using a voting-based classification scheme that integrates multiple signals and assigns taxonomic rank adaptively rather than using single best-hit assignments.