PROMALS3D

PROMALS3D constructs multiple sequence alignments of protein sequences by integrating sequence homologs, PSI-BLAST database searches, PSIPRED secondary-structure predictions, and available 3D structural information to improve alignment accuracy for divergent homologs.

Key Features:

• Integration of multiple data sources: Combines PSI-BLAST-derived homologs, PSIPRED secondary-structure predictions, available 3D structural information, and user-defined alignment constraints to inform alignments.
• Improved accuracy for distantly related sequences: Optimized for highly divergent homologs and reported to align a larger fraction of residues correctly at very low sequence identity (e.g., <10%).
• Advanced algorithmic components: Employs sequence database searches, secondary-structure predictions, a hidden Markov model that combines amino-acid and secondary-structure scoring, and probabilistic consistency-based scoring for progressive profile alignment.
• Comparative performance: Reported to outperform methods such as MUMMALS, ProbCons, MAFFT, SPEM, and HHalign, with largest gains for highly divergent homologs.
• Input and output formats and annotations: Accepts protein sequences and structures (FASTA and PDB formats) and produces alignments annotated with sequence grouping, predicted secondary structure, consensus sequences, and positional conservation, with intermediate search results available for inspection.

Scientific Applications:

• Homology detection: Improves detection of distant homologs by integrating profile and structural evidence.
• Structure modeling: Provides alignments suitable for comparative modeling and incorporation of 3D structural constraints.
• Function prediction: Facilitates inference of functional residues through conserved sequence and secondary-structure patterns.
• Phylogenetic analysis: Generates improved alignments for phylogenetic reconstruction of divergent protein families.
• Comparative protein analysis: Supports comparative analyses using annotated alignments and conservation metrics.

Methodology:

Performs sequence database searches (e.g., PSI-BLAST) to retrieve homologs, predicts secondary structure (e.g., PSIPRED), incorporates available 3D structural information, applies a hidden Markov model combining amino-acid and secondary-structure scoring, and uses probabilistic, consistency-based scoring to guide progressive profile alignment.