PATHNet: A Thermodynamic Path-Integral-Inspired Neural Network for Protein-Ligand Interaction Prediction
- 1Sravathi AI Technology Pvt Ltd
Abstract
Protein-ligand scoring functions must capture strongly nonlinear, target-dependent binding physics while remaining generalizable beyond their training data, a balance that purely analytic and purely data-driven models each struggle to achieve on their own. We present PATHNet, a physics-informed framework that learns this unknown functional relationship from molecular-mechanics descriptors using a non-monotonic piecewise-linear mapper with a bounded neural-network correction, trained through a variational, least-action-inspired objective and regulated by a renormalization-group-inspired spectral constraint, with model selection performed through multi-start basin optimization. Across seven DUD-E targets, PATHNet achieved the highest mean active inactive/decoy classification accuracy among the methods tested, outperforming three literature baselines representing distinct architectural families. These results support a physics-grounded, variationally constrained approach to neural-network scoring as a practical route toward more interpretable and transferable protein-ligand interaction models.
Related
- If the brain were so simple: Information-based drivers of intelligence and their interaction through an n-body-inspired framework
- Z-Hunt-DP: accelerating thermodynamic Z-DNA prediction with dynamic programming
- ReTIF: Granularity-Aware Multitask Interaction Routing for RNA-Compound Interaction Prediction and Binding-Site Localization
- Repurposing PeTriBERT for Protein Protein Interaction Prediction with Sequence Structure Fusion
- ProDTI: prior-guided positive-unlabeled learning for robust drug-target interaction prediction.
BibTeX
@misc{a2026pathnet,
title = {{PATHNet: A Thermodynamic Path-Integral-Inspired Neural Network for Protein-Ligand Interaction Prediction}},
author = {B. G. A. S. and A. Singh and S. Kanchan and N. M. Kulkarni},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.11.758399},
url = {https://www.biorxiv.org/content/10.64898/2026.10.11.758399v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - PATHNet: A Thermodynamic Path-Integral-Inspired Neural Network for Protein-Ligand Interaction Prediction
AU - B. G. A. S.
AU - A. Singh
AU - S. Kanchan
AU - N. M. Kulkarni
PY - 2026
DA - 2026/10/11
PB - bioRxiv
DO - 10.64898/2026.10.11.758399
UR - https://www.biorxiv.org/content/10.64898/2026.10.11.758399v1
AB - Protein-ligand scoring functions must capture strongly nonlinear, target-dependent binding physics while remaining generalizable beyond their training data, a balance that purely analytic and purely data-driven models each struggle to achieve on their own. We present PATHNet, a physics-informed framework that learns this unknown functional relationship from molecular-mechanics descriptors using a non-monotonic piecewise-linear mapper with a bounded neural-network correction, trained through a variational, least-action-inspired objective and regulated by a renormalization-group-inspired spectral constraint, with model selection performed through multi-start basin optimization. Across seven DUD-E targets, PATHNet achieved the highest mean active inactive/decoy classification accuracy among the methods tested, outperforming three literature baselines representing distinct architectural families. These results support a physics-grounded, variationally constrained approach to neural-network scoring as a practical route toward more interpretable and transferable protein-ligand interaction models.
ER -