SnaCleX: a browser-based structural biology workbench for protein-ligand interaction analysis, pocket prioritization, and exploratory docking

Interpreting how a small molecule sits in a protein usually means moving between a structure viewer, a command line docking program, and separate databases for chemistry and evolution. SnaCleX combines these steps in one browser session with no installation. A user loads a structure from the RCSB Protein Data Bank or uploads a PDB or mmCIF file, selects a bound molecule, and receives a classified list of atomic contacts with a per residue summary. A LIGSITE style grid search detects cavities and ranks them by a 0 to 100 druggability index, residues and pockets are scored by conservation from Pfam family alignments, and PubChem compounds can be docked into a chosen site by a grid based Monte Carlo search, with results cross referenced against ChEMBL. Every result records its method, parameters, random seed, and limitations. On trypsin with benzamidine, SnaCleX identified Asp189 as the key salt bridge partner and showed the S1 pocket to be the most conserved cavity. On the Astex Diverse (85 complexes) and PoseBusters Benchmark (428 complexes, 427 for the profile comparison) sets, the contact profiler recovered 96 to 97% of the residues PLIP reports in contact, with 61 to 65% precision. Rigid redocking of the crystal ligand conformer placed 39 to 40% of ligands within 2 Å (symmetry corrected RMSD) in a single run and 47 to 51% when the best scoring of three runs was kept, and a detected pocket lay within 6 Å of the ligand in 51% of structures. SnaCleX is therefore best used as a transparent tool for teaching and hypothesis generation rather than as a replacement for established docking engines. It is open source under the MIT License (DOI 10.5281/zenodo.23167411). Preprint, not peer reviewed. Version 1.