Optimal transport method for efficient modeling of extended peptide conformations

In this study, we expand our previous optimal-transport-based framework—initially developed to derive conformational distributions of tetrapeptides from dipeptide data in the Protein Data Bank (PDB)—to explore the conformations of longer peptide sequences. The limited availability of data for extended peptides in the PDB poses significant challenges in constructing optimized conformational landscapes for these sequences. To overcome this, we extend our methodology by incorporating the backbone torsional angle distributions of tetrapeptides from our prior work to model the conformational landscapes of longer peptides. This approach leverages a recursive framework that systematically combines the conformational distributions of shorter peptide fragments to build those of longer sequences, effectively managing the complex and high-dimensional nature of the peptide structures despite data limitations. A key innovation of our method lies in its ability to significantly reduce computational complexity: while conventional techniques face exponential growth in computational demands as peptide length increases, our approach scales more efficiently, enabling the practical exploration of extended peptide conformations. As a proof of concept, we have applied this methodology to peptides composed solely of alanine and glycine, including hexapeptides, octapeptides, decapeptides, and an 18-residue peptide, showcasing its capability to manage the intricate conformational space of extended peptides effectively and efficiently.