Connectivity-Aware Geometric Descriptors for Catalytic Site Alignment Across Non-Homologous Proteins

Abstract

Motivation: Enzymes with unrelated folds can share identical catalytic mechanisms through convergent evolution, yet aligning known catalytic sites across non-homologous proteins remains challenging. General-purpose structural aligners optimize global backbone similarity and sacrifice local catalytic geometry, while dedicated catalytic site tools achieve low recall on cross-fold matching.Results: This paper introduces CS-TOURIST, a catalytic site alignment method based on connectivity-aware octant descriptors. Each catalytic Cα neighborhood is encoded as 72 sub-slots combining spatial direction (8 octants × 3 radial zones) with sequence-connectivity class (local, medium, long-range), scored via class-presence matching robust to indels. Given annotated catalytic residues on both query and target, CS-TOURIST restricts its Smith-Waterman alignment to those residues and reports the geometric accuracy of the resulting superposition. Across 924 enzyme pairs from the Mechanism and Catalytic Site Atlas, CS-TOURIST achieves a median catalytic-site RMSD of 4.9 Å — 4.3× lower than the estimated random baseline of ~21 Å — with 19.4% of pairs achieving sub-2 Å RMSD and 51.2% under 5 Å. A negative control experiment replacing target catalytic residues with decoy residues (random and surface-exposed) demonstrates that the geometric scoring is not merely aligning arbitrary labelled sets: decoy RMSD is 2.2–2.7× higher than real catalytic RMSD (Wilcoxon p < 10⁻¹¹¹), with an overall AUC of 0.78–0.83 and 58–67% of pairs showing perfect separation. RMSD increases with evolutionary distance (same-superfamily median 4.3 Å → different-topology median 5.2 Å), confirming that the descriptor captures genuine geometric conservation rather than artefactual alignment. Availability: Python implementation and benchmark data are available at https://github.com/harkiolakis/cs-tourist and archived at https://doi.org/10.5281/zenodo.22331837.