SKiP: a Structure-oriented Kinetics Prediction suite for enzyme-substrate complexes

Predicting enzyme kinetic parameters remains a central challenge in enzyme engineering and biocatalysis. The current predictions are limited by lack of generalization and interpretability of existing sequence-centric predictors. We present SKiP (Structure-oriented Kinetics Prediction suite), a structure-aware framework that predicts turnover number (p k cat ) and Michaelis constant (p K m ) directly from enzyme-substrate complex structures, integrating global enzyme descriptors, local interaction fingerprints, active site descriptors, substrate physicochemical descriptors, and reaction-specific features within a single, mechanistically driven representation. Based on our recently curated dataset of enzyme-substrate three-dimensional structures, SKiP differentiates itself from the existing methods in three ways. First, it explicitly models holo enzyme context, including cofactor interactions where known, and pH-conditioned protonation states capturing environmental and active site determinants that are typically implicitly considered in sequence-only pipelines. Second, SKiP uses explicitly defined physicochemical descriptors of the enzyme, substrate, and reaction, facilitating analysis of the feature groups used by the model. Third, SKiP is more generalizable over unseen data. It can quantify aleatoric and epistemic uncertainty to distinguish high confidence from low confidence predictions at the individual data point level. Importantly, SKiP reproduces experimentally observed results in targeted validation studies. The model accurately captures pH-dependent catalytic behavior in a hyperthermophilic β-glucosidase and correctly discriminates between enantiomeric substrates in stereospecific lactate dehydrogenases, reproducing known active site interaction patterns. Collectively, SKiP provides a generalizable, intepretable, and uncertainty-aware method for structure-based enzyme kinetics prediction, enabling reliable prioritization for enzyme engineering and biocatalysis workflows. Scientific contribution SKiP is a structure-aware framework for predicting enzyme kinetic parameters from enzyme-substrate complex structures. Unlike sequence-centric approaches, it explicitly incorporates pH-conditioned holo-enzyme context, interpretable molecular interaction descriptors, and prediction uncertainty within a unified framework. SKiP demonstrates that explicit structural information improves generalizability across diverse enzymes, mutants, and substrates, providing a practical foundation for enzyme engineering and biocatalyst discovery.

Authors

Institutions

Publication Details

Journal
Journal of Cheminformatics
Published
2026-11-17
DOI
https://doi.org/10.1186/s13321-026-01333-z
Primary Topic
Protein Structure and Dynamics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

SKiP: a Structure-oriented Kinetics Prediction suite for enzyme-substrate complexes

Nishtha Pandey, Arijit Roy, Sowmya Ramaswamy Krishnan
3 links
Journal of Cheminformatics
Protein Structure and Dynamics
article

SKiP: a Structure-oriented Kinetics Prediction suite for enzyme-substrate complexes

Nishtha Pandey, Arijit Roy, Sowmya Ramaswamy Krishnan
article en

Abstract

Predicting enzyme kinetic parameters remains a central challenge in enzyme engineering and biocatalysis. The current predictions are limited by lack of generalization and interpretability of existing sequence-centric predictors. We present SKiP (Structure-oriented Kinetics Prediction suite), a structure-aware framework that predicts turnover number (p k cat ) and Michaelis constant (p K m ) directly from enzyme-substrate complex structures, integrating global enzyme descriptors, local interaction fingerprints, active site descriptors, substrate physicochemical descriptors, and reaction-specific features within a single, mechanistically driven representation. Based on our recently curated dataset of enzyme-substrate three-dimensional structures, SKiP differentiates itself from the existing methods in three ways. First, it explicitly models holo enzyme context, including cofactor interactions where known, and pH-conditioned protonation states capturing environmental and active site determinants that are typically implicitly considered in sequence-only pipelines. Second, SKiP uses explicitly defined physicochemical descriptors of the enzyme, substrate, and reaction, facilitating analysis of the feature groups used by the model. Third, SKiP is more generalizable over unseen data. It can quantify aleatoric and epistemic uncertainty to distinguish high confidence from low confidence predictions at the individual data point level. Importantly, SKiP reproduces experimentally observed results in targeted validation studies. The model accurately captures pH-dependent catalytic behavior in a hyperthermophilic β-glucosidase and correctly discriminates between enantiomeric substrates in stereospecific lactate dehydrogenases, reproducing known active site interaction patterns. Collectively, SKiP provides a generalizable, intepretable, and uncertainty-aware method for structure-based enzyme kinetics prediction, enabling reliable prioritization for enzyme engineering and biocatalysis workflows. Scientific contribution SKiP is a structure-aware framework for predicting enzyme kinetic parameters from enzyme-substrate complex structures. Unlike sequence-centric approaches, it explicitly incorporates pH-conditioned holo-enzyme context, interpretable molecular interaction descriptors, and prediction uncertainty within a unified framework. SKiP demonstrates that explicit structural information improves generalizability across diverse enzymes, mutants, and substrates, providing a practical foundation for enzyme engineering and biocatalyst discovery.

Journal of Cheminformatics
Tata Consultancy Services (India) (IN)
Protein Structure and Dynamics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon