OMAmer

OMAmer maps protein sequences to ancestral protein subfamilies using evolutionarily-informed k-mers to provide phylogeny-aware subfamily assignments for functional and evolutionary analyses.

Key Features:

• Evolutionarily-informed k-mers: Uses k-mers derived with phylogenetic information to represent sequence features used for mapping.
• Mapping to ancestral protein subfamilies: Assigns queries directly to ancestral protein subfamilies rather than only nearest-sequence hits.
• Rejection of non-homologous assignments: Detects and rejects non-homologous family-level assignments to reduce false-positive mappings.
• Improved subfamily-level categorization: Provides more accurate subfamily assignments than closest-sequence methods such as BLAST or DIAMOND, mitigating misassignments reported between 18% and 62% across datasets.
• Phylogeny-aware assignment: Incorporates gene phylogeny information to avoid over-specific subfamily assignments caused by neglecting phylogeny.
• Avoidance of gene tree inference: Achieves improved assignment accuracy without requiring computationally expensive gene tree inference.
• Computational efficiency: Designed for scalable analysis in large-scale genomic studies where efficiency and accuracy are important.

Scientific Applications:

• Functional genomics: Enables more reliable functional annotation through accurate protein subfamily assignments.
• Comparative genomics: Supports comparative analyses by mapping proteins across species to conserved ancestral subfamilies.
• Evolutionary genomics: Facilitates evolutionary inference by providing phylogeny-aware subfamily categorizations.
• Large-scale genomic studies: Applies to high-throughput datasets where scalable, accurate subfamily assignment is needed.

Methodology:

Uses evolutionarily-informed k-mers to map sequences to ancestral protein subfamilies, leverages phylogenetic information to reject non-homologous family-level assignments and improve subfamily-level categorization relative to BLAST or DIAMOND, and does so without requiring gene tree inference.