Argot2

Argot2 annotates nucleic and protein sequences with Gene Ontology (GO) terms to provide automated functional predictions for genomic and proteomic datasets generated by next-generation sequencing technologies.

Key Features:

• Integrated approach: Argot2 clusters GO terms based on semantic similarity and applies a weighting scheme that assesses hits sharing biological features with the target sequence to improve functional inference precision.
• Data processing and input: Accepts FASTA input and can process datasets ranging from small collections to entire genomes using BLAST and HMMER searches against UniProtKB and Pfam.
• Semantic similarity and weighting: Retrieves GO terms from the UniProtKB-GOA database and weights them using e-values from BLAST and HMMER combined with GO-term semantic similarity to produce association scores.
• Enhanced algorithm: Implements a fully rewritten algorithm (successor to Argot) that improves accuracy and computational efficiency, enabling high-precision, high-recall annotation of complete genomes.
• Benchmarking and validation: Benchmarked on 10,000 Saccharomyces cerevisiae protein sequences and other datasets, demonstrating improved sensitivity, specificity, and coverage.
• Comparative performance: Shows superior performance metrics in comparisons with Blast2GO.

Scientific Applications:

• Genomic annotation: Annotation of entire genomes and large-scale genomic projects.
• Functional inference: Genome-scale prediction of gene and protein functions using GO terms to aid interpretation of gene functions and interactions.
• Applied research: Applied in in-house genome projects such as grape and apple to generate functional annotations.

Methodology:

Argot2 performs BLAST and HMMER searches against UniProtKB and Pfam, retrieves GO annotations from UniProtKB-GOA, clusters GO terms by semantic similarity, weights GO-term associations using BLAST/HMMER e-values and semantic similarity relations, and assigns GO-term scores using its rewritten Argot algorithm.