SIMAP

SIMAP provides pre-calculated protein sequence similarities and associated annotations to enable large-scale protein similarity analysis and functional annotation.

Key Features:

• Pre-calculated sequence similarities: A comprehensive matrix of pre-computed pairwise protein similarities interconnecting the known protein sequence universe.
• Pre-computed protein features and domains: Protein features and domain annotations are provided for sequences across the database.
• Similarity clusters: Pre-defined similarity clusters group related protein sequences.
• Functional annotations: Functional annotations are provided, including those derived by the BLAST2GO algorithm.
• Coverage and scale: As of September 2013, the resource contains over 163 million proteins representing approximately 70 million non-redundant sequences.
• Data integration: Integrates data from all major public protein databases and consistently re-annotated metagenomes.
• Alignment methods: Employs sensitive FASTA search heuristics and Smith-Waterman alignments for similarity assessment.
• Domain models: Uses InterPro database models for protein domain annotation.
• All-against-all matrices: Ability to generate all-against-all protein sequence similarity matrices for specified protein collections.
• Protein similarity networks: Provides extraction of protein similarity networks for analysis of relationships and functions.

Scientific Applications:

• Protein sequence exploration: Comparative exploration of protein sequences across the global protein sequence universe.
• Functional annotation: Assignment and propagation of functional terms using BLAST2GO and InterPro-derived domain information.
• Comparative analyses: Large-scale all-against-all similarity analyses for project-specific protein collections.
• Network analysis: Construction and analysis of protein similarity networks to investigate protein relationships and functional modules.
• Non-redundant sequence mining: Analysis and mining of non-redundant sequence space at large scale.

Methodology:

Pre-calculation of sequence similarities using sensitive FASTA search heuristics and Smith-Waterman alignments; annotation with InterPro domain models and BLAST2GO functional annotation; generation of all-against-all similarity matrices, similarity clustering, and extraction of protein similarity networks.