SLiMFold Pipeline

Have you noticed two or more short peptide sequences that bind to the same target protein in nearly identical ways? If so, you may have discovered a novel short linear motif (SLiM)! SLiMs are brief amino acid segments that often mediate critical protein–protein interactions. Identifying and characterizing these motifs can be challenging, but that’s precisely where SLiMFold comes in. SLiMFold is a modular pipeline designed to systematically uncover, analyze, and refine potential SLiMs using a powerful combination of sequence- and structure-based bioinformatics tools. From an initial set of suspected motif-containing sequences, SLiMFold will: Generate Position-Specific Scoring Matrices (PSSMs) from your input motifs. Search Large Proteomes to find novel motif occurrences, while filtering based on predicted secondary structure, disorder, and binding regions. It also builds multiple sequence alignments (MSAs) and prepares peptide–bait pairs for downstream prediction (Prerun.ipynb). Perform structural predictions of motif–target complexes using AlphaFold2 via a modified ColabFold interface (ColabFold_looped.ipynb) Conduct robust postanalysis to extract structural metrics, visualize results, and cluster similar predictions to identify distinct motif classes (Postanalysis.ipynb). Whether you're validating a new SLiM or exploring uncharted motif landscapes, SLiMFold streamlines the process and delivers predictions ready for experimental follow-up. Below you will find instructions for installing dependencies, running each notebook, and interpreting the results. (Note that for very large bait proteins (> 2,950 residues), a H100 GPU might be required, which is not directly accessible within ColabFold. Also be aware that the database for MSA generation, used in this pipeline (UniRef90), may be suboptimal for viral proteins, synthetic constructs, or highly divergent sequences, due to limited homologous coverage in these cases.).