ASM-Clust

ASM-Clust performs de novo classification of large, functionally diverse protein families by clustering an alignment score matrix to improve functional interpretation of genome-derived protein sequences.

Key Features:

• De Novo Classification: Performs de novo classification of large protein superfamilies from sequence data, applicable to proteins derived from environmental samples.
• Alignment Score Matrix Clustering: Generates and clusters an alignment score matrix created by aligning all sequences within a family to a small subset of sequences to resolve functional subgroups.
• Implementation: Implemented in bash with two helper Perl scripts and processes a single protein FASTA file via the primary script ASM_clust.sh.
• Evaluation: Demonstrated on the enolase family using annotations documented in the Structure Function Linkage Database.

Scientific Applications:

• Environmental and metagenomic protein classification: Classifies proteins from genomes derived from environmental samples where only sequence information is available.
• Functional annotation refinement: Identifies functionally distinct subgroups to reduce propagation of misannotation from homology-based annotation transfer.
• Comparative and ecological inference: Supports inference of organism physiology and ecology from genome sequences by improving functional predictions across diverse datasets.

Methodology:

Clusters an alignment score matrix generated by aligning all sequences within a family to a small subset of the data; implemented in bash with two Perl helper scripts; processes input from a single protein FASTA file using ASM_clust.sh.