Hidden Markov models in computational biology. Applications to protein modeling

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Bioinformatics

Background:

• Protein sequence analysis is crucial for understanding protein function and evolution.
• Accurate multiple sequence alignment and database searching are fundamental tasks in bioinformatics.
• Existing methods for protein family identification and domain analysis have limitations.

Purpose of the Study:

• To apply Hidden Markov Models (HMMs) for statistical modeling, database searching, and multiple sequence alignment of protein families and domains.
• To evaluate the performance of HMMs in comparison to other established methods like PROSITE and PROFILESEARCH.
• To identify potential functional motifs, such as the EF-hand calcium binding motif, within specific protein structures.

Main Methods:

• Estimation of HMM parameters from unaligned training sequences.
• Utilizing trained HMMs for multiple sequence alignment of training data.
• Searching the SWISS-PROT database using HMMs to identify homologous sequences and domains.
• Performing discrimination tests to assess the accuracy of HMMs in distinguishing protein families.

Main Results:

• HMMs generated high-quality multiple sequence alignments comparable to structure-based methods.
• HMMs accurately distinguished members of protein families (globin, kinase, EF-hand) from non-members.
• HMMs outperformed PROSITE and showed a slight advantage over PROFILESEARCH in discrimination tests, with lower false positive and negative rates.
• A conserved EF-hand calcium binding motif was identified in the alpha-1 subunit of L-type calcium channels.

Conclusions:

• HMMs are a powerful and accurate tool for protein family analysis, sequence alignment, and database searching, even when trained on unaligned sequences.
• HMMs offer improved performance over existing methods like PROSITE and PROFILESEARCH for identifying protein families and domains.
• The identified EF-hand motif suggests functional roles in the intracellular region of L-type calcium channels, impacting excitation-contraction coupling.