Prediction of protein-ligand binding modes is a key step in computer-aided drug design, since incorrect pose prediction inevitably limits the achievable accuracy of downstream physics-based affinity predictions. The ASAP-Polaris-OpenADMET antiviral ...

Parameter Refinement Engine for Smirnoff Training / Optimisation. Train bespoke SMIRNOFF force fields quickly using a machine learning potential - cole-group/presto

Partial atomic charges are a fundamental component underlying classical molecular simulations, but assigning charges remains a computational bottleneck; many common methods rely on quantum mechanical ...

Postdoctoral Fellowships offer researchers holding a PhD the opportunity to acquire new skills through advanced training and international, interdisciplinary, and inter-sectoral mobility.

The increasing importance and predictive power of modern molecular modeling, driven by physics- and machine-learning-based methods, necessitates a new collaborative architecture to replace the isolate...

This release adds openff-2.3.0.offxml and openff_unconstrained-2.3.0.offxml. Sage 2.3.0 is the first OpenFF force field to use the AshGC neural network charge model to assign charges. Both vdW para...

We report a new charge model and a new general small molecule force field. Here, we address the development and benchmarking of both the Open Force Field (OpenFF) AshGC charge model, as well as the Sa...

This link will take you to a page that’s not on LinkedIn

Common methods for assigning atom-centered partial charges in computational chemistry, such as RESP and AM1-BCC, rely on quantum mechanical or semiempirical calculations of the molecule of interest, which are expensive to compute and dependent on the choice of input molecular conformer(s). Graph neural network (GNN) based continuous atom embeddings have been shown to be a fast and flexible solution for partial charge assignment, but those developed so far for condensed phase modeling have usually been trained to reproduce AM1-BCC charges, which themselves seek to reproduce the HF/6-31G(d) molecular electrostatic potential. Here, we investigate the suitability of various common charge assignment schemes, including ESP and atoms-in-molecule (AIM) based approaches, as training targets for new GNN-based charge models. We show that the strengths of both approaches can be combined by cotraining GNN models to AIM charges and molecular dipoles and electrostatic potentials. We collect a data set of quantum mechanical AIM properties computed at a high level of theory (ωB97X-D/def2-tzvpp), in both vacuum and implicit solvent, and train new GNN charge models to each. Charges can be scaled between the vacuum and solvated charge sets, and combined with Lennard-Jones parameters optimized using the Open Force Field infrastructure, to yield force fields that are suitably polarized for condensed phase modeling. We further demonstrate that the charge models may be applied to explore electrostatics-driven structure–activity relationships in medicinal chemistry. The charge models are freely available at: https://github.com/cole-group/nagl-mbis/.

Hybrid ML/MM approaches that combine machine learning (ML) potentials with molecular mechanics (MM) potentials offer a promising balance between computational cost and accuracy. Most ML/MM simulations reported to date employ mechanical embedding schemes, and rely on Lennard-Jones and Coulomb potentials to model intermolecular interactions between the ML and MM regions. A promising approach to improving ML/MM schemes is to use electrostatic embedding, where polarization effects on the ML region by the MM region are explicitly incorporated. The electrostatic machine learning embedding (EMLE) method has been developed for this purpose. Here, we compute absolute hydration free energies for a set of small organic molecules to derive robust methodologies for training EMLE models using quantum mechanical data. We establish protocols for fine-tuning the static and induced components of electrostatic interactions and evaluate the accuracy limits of fitting these components to first-principles calculations. We also introduce an empirical adjustment to enhance agreement with experimental results, strengthening the competitiveness of ML/MM simulations relative to state-of-the-art methods. Overall, our findings provide valuable insights into the challenges and opportunities of electrostatic embedding ML/MM simulations, and offer strategies for achieving robust modeling of classes of drug-like molecules where the accuracy of conventional MM force fields fall short.