Simultaneous alignment and clustering of peptide data using a Gibbs sampling approach.

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Biology

Background:

• Peptide-binding proteins are crucial for cellular signaling.
• High-throughput technologies enable large-scale peptide-binding specificity studies.
• Analyzing complex peptide datasets with multiple motifs is challenging.

Purpose of the Study:

• To develop an algorithm for identifying multiple receptor binding motifs in peptide datasets.
• To address the complexity of analyzing large peptide libraries with varying motif locations and multiple specificities.

Main Methods:

• A novel algorithm based on Gibbs sampling.
• Simultaneous alignment and clustering of peptide data.
• Application to datasets of varying complexity, including fixed-length and variable-length peptides.

Main Results:

• The algorithm successfully identifies multiple specificities in peptide data.
• Demonstrated de-convolution of binding motifs in diverse datasets.
• Successfully analyzed mixtures of binders for MHC class I/II alleles, SH3 domains, and HLA-A*02:01 sub-specificities.

Conclusions:

• The Gibbs clustering method provides an effective solution for analyzing complex peptide-binding data.
• The algorithm facilitates a deeper understanding of protein-peptide interactions.
• The method is accessible as a web server for broader scientific use.