Skip to content

Tractable and Thermodynamically Consistent pKa Prediction via First-Dissociation Aggregation

Oct 2026 · bioRxiv · 0 citations · 40 references
Biology

Abstract

Ionization governs molecular behavior, yet predicting aqueous pKa accurately and tractably remains a fundamental challenge. Rigorous ensemble methods require enumerating a protonation-state space that grows exponentially with the number of ionizable sites, while fast graph predictors return numbers without thermodynamic consistency. Here we derive an exact thermodynamic identity showing that the macroscopic first-dissociation constant is fully determined by protonated-microstate populations and unique first-deprotonation events, eliminating the deprotonated ensemble algebraically and reducing aggregation to O(n+E) operations. We implement this identity in DTi-pKa, a dual-head graph neural network whose free-energy and dissociation heads are coupled by thermodynamic constraints. On four external benchmarks (355 molecules), DTi-pKa achieves a pooled MAE of 0.5665 pKa units, with 0.6285 on the 169 records lacking training-cache matches, while providing site-resolved micro-pKa values and protonation-state populations. Controlled ablations show that constraint placement at the level of the reported macroscopic equilibrium matters more than microscopic label precision—a design principle transferable to physics-informed machine learning beyond chemistry. Closure diagnostics expose numerical self-consistency as an open frontier, which we report transparently. By replacing an intractable computation with a physical identity, DTi-pKa unifies macroscopic accuracy and microscopic interpretability in a single framework.

Read PDF

Similar papers

Open access Oct 2026

Determining the Partial Molar Volume of the Hydrated Electron from First-Principles Simulated Solution Volumes.

The microscopic solvation structure of the hydrated electron remains difficult to characterize directly despite extensive experimental and theoretical efforts. The partial molar volume,V¯e-, provides an indirect thermodynamic probe, but quantitative agreement with experiment is challenging due to the strong dependence...

Meng-Xu Li, Ruo-Ting Zhao, Chen-Yu Zhang et al. · 0 citations
Open access Sep 2026

Bridging the accuracy-speed divide in reactive molecular dynamics with QuantaMind MD

QuantaMind is presented, a machine learning force field framework that attains density functional theory–level accuracy, enabling fully reactive simulations of complex molecular systems, and establishes a paradigm for long-timescale, fully reactive molecular simulation.

Song Xia, De-Qiang Zhang, Xue-Dong Shang et al. · 0 citations
Open access Oct 2026

Conformational heterogeneity and vibrational signatures of dAMP in aqueous solution revealed by high-accuracy machine-learning potentials

Nucleotides are the fundamental building blocks of life, yet many aspects of their behavior and interactions in water remain poorly understood. Here, we investigate the structural and vibrational properties of $${2}^{{\prime} }$$ 2 ′ -deoxyadenosine $${5}^{{\prime} }$$ 5...

Laurie Stevens, R. Martina, Indrani Bhattacharya et al. · 0 citations
Open access Oct 2026

Halide Identity Shapes Ion Pairing and Solvent Dynamics in Aqueous Lithium Halide Electrolytes.

Predicting whether electrolytes associate or dissociate in water is challenging because direct ion attraction, hydration, solvent caging, and collective reorganization operate on comparable free-energy scales, and subtle changes in their balance can reverse the preferred ion-pairing state. Here, we develop machine-lear...

Dilshan Karunachandra, Subhajit Acharya, S. Dasgupta · 0 citations
Preprint Oct 2026

Learning atomic interactions for the thermodynamics of liquid mixtures

Predictive models for activity coefficients, which describe the non-ideality of liquid mixtures, are foundational for process design and optimization. Group-contribution methods, such as UNIFAC, are computationally cheap and well established in process simulators. Modern machine learning models, such as HANNA, offer hi...

M. Hoffmann, S. Schmitt, H. Hasse et al. · 0 citations
#machine learning Preprint Oct 2026

Solvating machine-learning interatomic potentials for solution-phase chemistry

Most machine-learning interatomic potentials (MLIPs) are trained on gas-phase data, which limits their use for chemistry in solution. Here, we introduce AnySolv, a transferable implicit solvation potential that provides solvent-dependent energy and force corrections to existing gas-phase potentials without retraining t...

Anton Z. Ni, J. Lee · 0 citations

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.