Reduced alphabet motif methodology for GPCR annotation

Area of Science:

• Computational biology
• Bioinformatics
• Genomics

Background:

• G-protein coupled receptors (GPCRs) classification is computationally challenging.
• Existing methods like machine learning and sequence motif analysis have limitations such as class imbalance, lack of multi-class classification, and over-reliance on primary sequence similarity.

Purpose of the Study:

• To develop and evaluate a novel motif discovery and application methodology for GPCR classification.
• To overcome the limitations of existing machine learning and sequence similarity-based approaches.
• To accurately classify GPCRs, including orphan receptors, and predict their families and subfamilies.

Main Methods:

• Utilizing a reduced protein functional alphabet representation where functionally similar residues share symbols.
• Generating regular expression motifs via ClustalW based multiple sequence alignment with an identity matrix.
• Employing this method for sequences without necessarily sharing common ancestry, avoiding evolutionary matrices like BLOSUM and PAM.

Main Results:

• The reduced alphabet motifs accurately classify known GPCR proteins, achieving results comparable to PRINTS and PROSITE.
• Demonstrated high accuracy with no false negatives and few false positives for well-known GPCRs in SWISSPROT.
• Successfully classified most known GPCR classes, even with limited representative sequences, and predicted multiple classes for certain sequences.
• Annotated 695 orphan receptors, identifying 121 belonging to Family A.

Conclusions:

• The proposed methodology offers an effective and versatile approach for GPCR classification.
• It overcomes key limitations of traditional machine learning and sequence similarity methods.
• The developed web interface provides a practical tool for predicting GPCR families and subfamilies.