DOUTfinder

DOUTfinder identifies protein domains in remotely related (low-similarity) sequences to detect subsignificant domain similarities and prioritize likely homologous domains for studies of protein function and evolution.

Key Features:

• Sequence input and PSI-BLAST expansion: accepts user-supplied sequence sets or expands a single query with PSI-BLAST to assemble a sequence family for analysis.
• Domain annotation with Pfam and SMART: analyzes the assembled sequence family against Pfam and SMART to detect known domains.
• Subsignificant similarity detection: identifies domain similarities below conventional significance thresholds (subsignificant hits) across the sequence family.
• Sum-score ranking: aggregates multiple subthreshold hits into a sum-score to rank domains by support across the query set.
• Post-filtering of subsignificant hits: applies a post-filtering step to refine subsignificant domain hits and highlight non-trivial domain relationships.
• Established cut-offs for homology suggestion: employs established cut-offs to suggest likely homologous domains.

Scientific Applications:

• Protein evolution inference: detection of distant homologous domains to inform evolutionary relationships among protein families.
• Functional annotation of divergent sequences: support assignment of putative domain functions in sequences with low overall similarity.
• Discovery of non-trivial domain relationships: uncover domain connections overlooked by conventional significance thresholds to study protein family diversification.

Methodology:

Collect sequence data (user-supplied or PSI-BLAST expansion), analyze the sequence family for known domains using Pfam and SMART, evaluate subsignificant similarities, rank domains using a sum-score function, apply a post-filtering procedure to subsignificant hits, and use established cut-offs to suggest likely homologous domains.