Argot 2.5

Argot 2.5 predicts protein function by inferring Gene Ontology (GO) terms from sequence and domain evidence using semantic similarity, Pfam hit weighting, a clustered UniProt reference set, and taxonomic constraints.

Key Features:

• Semantic Similarity Grouping: Groups Gene Ontology (GO) terms based on semantic similarity to improve prediction accuracy.
• Clustered UniProt and Pfam Weighting: Refines input data selection from sequence similarity searches using a clustered version of the UniProt database and optimizes identification and weighting of Pfam hits.
• Taxonomic Constraints (FunTaxIS): Applies taxonomic constraints derived from FunTaxIS that extend Gene Ontology consortium rules to filter annotations incompatible with the query species.
• Benchmarking and Comparative Evaluation: Evaluates performance against two independent benchmarks and compares results to Argot2, PANNZER, and a BLAST-based baseline.
• High-throughput Annotation: Processes thousands of sequences to support large-scale functional annotation projects.

Scientific Applications:

• High-throughput protein annotation: Provides GO-term annotations for large protein sets to enable broad functional characterization.
• Genomic and proteomic projects: Facilitates functional annotation in genome and proteome studies.
• Functional genomics and systems biology: Supports functional genomics and systems-level analyses by supplying GO annotations for pathway and interaction studies.
• Pathway and process inference: Aids interpretation of biological processes, pathways, and molecular interactions through GO-based functional assignments.

Methodology:

Groups GO terms by semantic similarity; selects and weights sequence and Pfam hits using a clustered UniProt database; applies taxonomic filters from FunTaxIS that extend Gene Ontology consortium rules; and benchmarks predictions against two independent datasets and methods including Argot2, PANNZER, and a BLAST-based baseline.