Prediction of functional specificity determinants from protein sequences using log-likelihood ratios.

Area of Science:

• Computational Biology
• Bioinformatics
• Protein Science

Background:

• Existing methods for predicting protein functional specificity determinants often require pre-defined subgroups, which are difficult to establish.
• Manual subgroup definition is limited by experimental data, and computational partitioning can be unreliable.

Purpose of the Study:

• To introduce SPEL (specificity positions by evolutionary likelihood), a new computational method for identifying functional specificity determinants in protein families.
• To develop a method that does not require prior subgroup definition, using only multiple sequence alignments as input.

Main Methods:

• SPEL assigns a P-value to each position in a protein family's multiple sequence alignment.
• It reconstructs an evolutionary tree and uses evolutionary simulations to estimate P-values based on amino acid distribution (evolutionary log-likelihood).
• The method was validated by analyzing the LacI/PurR and G protein alpha subunit families and comparing it with existing methods.

Main Results:

• SPEL successfully identifies positions likely to be functional specificity determinants.
• The method's performance was comparable to or better than existing approaches like evolutionary trace and mutual information.
• Comparisons were made on diverse protein families, including those with known ligand-bound structures.

Conclusions:

• SPEL provides an effective, subgroup-independent approach for detecting functional specificity determinants.
• The method offers a valuable tool for protein family analysis and understanding protein function.
• SPEL is available for non-commercial use.