Prediction of transmembrane segments in proteins utilising multiple sequence alignments

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Transmembrane segments are crucial for protein function and localization within cell membranes.
• Accurate prediction of transmembrane segments is essential for understanding protein structure-function relationships.

Purpose of the Study:

• To develop and validate a novel computational method for predicting transmembrane segments from multiply aligned amino acid sequences.
• To enhance the accuracy of transmembrane segment prediction compared to existing single-sequence methods.

Main Methods:

• Utilized two sets of propensity values (hydrophobic middle, terminal regions) for transmembrane segment prediction.
• Employed weighted average propensity calculations based on sequence dissimilarity within alignments.
• Incorporated segment elongation, stop signals, length constraints (15-29 residues), and charge corrections.

Main Results:

• The method demonstrated higher success rates than single-sequence predictions.
• In a test set of 28 families (126 transmembrane segments), only five spans were missed or were false positives.
• Successfully applied to predict transmembrane segments in various protein families with limited or contradictory data.

Conclusions:

• The developed method offers a robust and accurate approach for predicting transmembrane segments from multiple sequence alignments.
• This tool is valuable for analyzing protein families with sparse or uncertain transmembrane topology data.
• The improved prediction accuracy aids in the functional and structural characterization of membrane proteins.