CMCharles J. McGill

Charles J. McGill

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Charles J. McGill has a total of 46 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Machine learning guided optimization of an oral microemulsion system: a Bayesian optimization approach

    by Madeline Gunawardena, Bao Chau, Hayden Nothacker, Rudra Pangeni, Thomas Roper, Qingguo Xu, Charles McGill

    Oral microemulsions are one drug delivery system often implemented to improve intestinal permeability and oral bioavailability of poorly water-soluble drugs. They also present several practical advantages including high patient compliance and simplified manufacturing methods which contribute to their promise as marketable drug products. Despite these advantages, however, the microemulsion formulation development process is extremely time consuming and resource intensive, typically involving extensive screening of components and excessive preliminary trials in order to achieve stable formulations. As a result, MEs are often suboptimal in their therapeutic performance, and there is an unmet need for improved methods to streamline microemulsion formulation design. In this work, a batch Bayesian Optimization strategy was used to design a subset of unique in-specification microemulsions with highly optimized physicochemical properties in five iterations containing batches of five experiments. A two-phase modeling approach was developed to achieve this goal and allowed for navigating a complex experimental design space, including multiple oils, surfactants, cosurfactants, and processing parameters with an initial training dataset consisting of 22 experiments. As a result of this study, five high-performing blank microemulsions were identified, and four of them exhibited physical stability upon storage for up to 30 days. Further, when loaded with two different model drug candidates, three of the microemulsions achieved high drug loading, acceptable stability and improved in-vitro permeability, highlighting their promise for potential translation to later stages of development.

  • Open Access

    Machine Learning Guided Optimization of an Oral Microemulsion System: A Bayesian Optimization Approach

    by Madeline Gunawardena, Bao Chau, Hayden Nothacker, Rudra Pangeni, Thomas Roper, Qingguo Xu, Charles McGill

    Oral microemulsions are one drug delivery system often implemented to improve intestinal permeability and oral bioavailability of poorly-water soluble drugs. They also present several practical advantages including high patient compliance and simplified manufacturing methods which contribute to their promise as marketable drug products. Despite these advantages, however, the microemulsion formulation development process is extremely time consuming and resource intensive, typically involving extensive screening of components and excessive preliminary trials in order to achieve stable formulations. As a result, MEs are often suboptimal in their therapeutic performance, and there is an unmet need for improved methods to streamline microemulsion formulation design. In this work, a batch Bayesian Optimization strategy was used to design a subset of unique in-specification microemulsions with highly optimized physicochemical properties in five iterations containing batches of five experiments. A two-phase modeling approach was developed to achieve this goal and allowed for navigating a complex experimental design space, including multiple oils, surfactants, cosurfactants, and processing parameters with a training dataset consisting of 22 experiments. As a result of this study, five high-performing blank microemulsions were identified, and four of them exhibited physical stability upon storage for up to 30 days. Further, when loaded with two different model drug candidates, three of the microemulsions achieved high drug loading, acceptable stability and improved in-vitro permeability, highlighting their promise for potential translation to later stages of development.

  • Open Access

    EquiNet: Modeling Vapor-Liquid Equilibrium Using Simultaneous Neural-Network Predictions for Activity Coefficient and Vapor Pressure

    by M. Zaher Alam, Abdullah Alqallaf, Charles McGill

    In this work, we present EquiNet, a neural network for predicting vapor–liquid equilibrium (VLE) in novel binary mixtures through direct estimation of activity coefficients and vapor pressures. The model embeds a classic excess-Gibbs free energy formulation, ensuring Gibbs–Duhem consistency on all predicted activity coefficients. Given molecular SMILES, liquid-phase composition, and temperature, EquiNet predicts vapor composition and pressure, enabling end-to-end training on experimental data without requiring precomputed activity coefficients. Across the UNIFAC-scope VLE dataset (covering 55,470 binary datapoints), EquiNet achieves a log10 Pressure (Pa) RMSE of 0.0659, outperforming UNIFAC (0.087) and the HANNA machine-learning model (0.101) under the same molecular-scope assumptions. For evaluation on previously unseen chemical systems, we follow a novel-mixture cross-validation protocol that demonstrated strong transferability across diverse functional groups and molecular classes. On the public Jaubert benchmark, restricted to UNIFAC-compatible pairs (2,099 points), EquiNet attains a comparable 0.03436 log10 Pressure (Pa) RMSE, substantially below HANNA (0.26743) and UNIFAC (1.55616). This work employs a modified Raoult’s law approach without vapor-phase fugacity or Poynting corrections and is limited to neutral, organic, non-reactive binary mixtures composed of supported elements. The method does not detect mixture criticality, and any supercritical examples should be interpreted qualitatively.


  • Open Access

    Transitory Sensitivity in Automatic Chemical Kinetic Mechanism Analysis

    by Matthew S. Johnson, Charles J. McGill, William H. Green

    Detailed chemical kinetic mechanisms are necessary for resolving many important chemical processes. As the field has advanced, kineticists have become interested in constructing progressively larger kinetic mechanisms to model increasingly complex chemical processes. These large kinetic mechanisms prove incredibly difficult to refine and time-consuming to interpret. Traditional sensitivity analysis on a large mechanism can range from inconvenient to practically impossible without special techniques to reduce the computational cost. We first present a new time-local sensitivity analysis we term transitory sensitivity analysis. Transitory sensitivity analysis is demonstrated in an example to accurately identify traditionally sensitive reactions at an 18,000x speed up over traditional sensitivities. By fusing transitory sensitivity analysis with more traditional time-local branching, pathway and cluster analyses, we develop an algorithm for efficient automatic mechanism analysis. This automatic mechanism analysis at a time point is able to identify the reactions a target is most sensitive to using transitory sensitivity analysis and then propose hypotheses why the reaction might be sensitive using branching, pathway, and cluster analyses. We implement these algorithms within the Reaction Mechanism Simulator (RMS) package, which enables us to report the the automatic mechanism analysis results in highly readable text format and in molecular flux diagrams.

  • Top Cited Papers
    Open Access

    Chemprop: A Machine Learning Package for Chemical Property Prediction

    by Esther Heid, Kevin P. Greenman, Yunsie Chung, Shih-Cheng Li, David E. Graff, Florence H. Vermeire, Haoyang Wu, William H. Green, Charles J. McGill

    Deep learning has become a powerful and frequently employed tool for the prediction of molecular properties, thus creating a need for open-source and versatile software solutions that can be operated by nonexperts. Among the current approaches, directed message-passing neural networks (D-MPNNs) have proven to perform well on a variety of property prediction tasks. The software package Chemprop implements the D-MPNN architecture and offers simple, easy, and fast access to machine-learned molecular properties. Compared to its initial version, we present a multitude of new Chemprop functionalities such as the support of multimolecule properties, reactions, atom/bond-level properties, and spectra. Further, we incorporate various uncertainty quantification and calibration methods along with related metrics as well as pretraining and transfer learning workflows, improved hyperparameter optimization, and other customization options concerning loss functions or atom/bond features. We benchmark D-MPNN models trained using Chemprop with the new reaction, atom-level, and spectra functionality on a variety of property prediction data sets, including MoleculeNet and SAMPL, and observe state-of-the-art performance on the prediction of water-octanol partition coefficients, reaction barrier heights, atomic partial charges, and absorption spectra. Chemprop enables out-of-the-box training of D-MPNN models for a variety of problem settings in fast, user-friendly, and open-source software.


  • Open Access

    Chemprop: A Machine Learning Package for Chemical Property Prediction

    by Esther Heid, Kevin P. Greenman, Yunsie Chung, Shih-Cheng Li, David E. Graff, Florence H. Vermeire, Haoyang Wu, William H. Green, Charles J. McGill

    Deep learning has become a powerful and frequently employed tool for the prediction of molecular properties, thus creating a need for open-source and versatile software solutions that can be operated by non-experts. Among current approaches, directed message-passing neural networks (D-MPNNs) have proven to perform well on a variety of property prediction tasks. The software package Chemprop implements the D-MPNN architecture, and offers simple, easy, and fast access to machine-learned molecular properties. Compared to its initial version, we present a multitude of new Chemprop functionalities such as the support of multi-molecule properties, reactions, atom/bond-level properties, and spectra. Further, we incorporate various uncertainty quantification and calibration methods along with related metrics, as well as pretraining and transfer learning workflows, improved hyperparameter optimization, and other customization options concerning loss functions or atom/bond features. We benchmark D-MPNN models trained using Chemprop with the new reaction, atom-level and spectra functionality on a variety of property prediction datasets, including MoleculeNet and SAMPL, and observe state-of-the-art performance on the prediction of water-octanol partition coefficients, reaction barrier heights, atomic partial charges, and absorption spectra. Chemprop enables out-of-the-box training of D-MPNN models for a variety of problem settings in a fast, user-friendly, and open-source software.


  • Open Access

    Characterizing Uncertainty in Machine Learning for Chemistry

    by Esther Heid, Charles J. McGill, Florence H. Vermeire, William H. Green

    Characterizing uncertainty in machine learning models has recently gained interest in the context of machine learning reliability, robustness, safety, and active learning. Here, we separate the total uncertainty into contributions from noise in the data (aleatoric) and shortcomings of the model (epistemic), further dividing epistemic uncertainty into model bias and variance contributions. We systematically address the influence of noise, model bias, and model variance in the context of chemical property predictions, where the diverse nature of target properties and the vast chemical chemical space give rise to many different distinct sources of prediction error. We demonstrate that different sources of error can each be significant in different contexts and must be individually addressed during model development. Through controlled experiments on data sets of molecular properties, we show important trends in model performance associated with the level of noise in the data set, size of the data set, model architecture, molecule representation, ensemble size, and data set splitting. In particular, we show that 1) noise in the test set can limit a model's observed performance when the actual performance is much better, 2) using size-extensive model aggregation structures is crucial for extensive property prediction, and 3) ensembling is a reliable tool for uncertainty quantification and improvement specifically for the contribution of model variance. We develop general guidelines on how to improve an underperforming model when falling into different uncertainty contexts.


  • Open Access

    Characterizing Uncertainty in Machine Learning for Chemistry

    by Esther Heid, Charles J. McGill, Florence H. Vermeire, William H. Green

    Characterizing uncertainty in machine learning models has recently gained interest in the context of machine learning reliability, robustness, safety, and active learning. Here, we separate the total uncertainty into contributions from noise in the data (aleatoric) and shortcomings of the model (epistemic), further dividing epistemic uncertainty into model bias and variance contributions. We systematically address the influence of noise, model bias, and model variance in the context of chemical property predictions, where the diverse nature of target properties and the vast chemical chemical space give rise to many different distinct sources of prediction error. We demonstrate that different sources of error can each be significant in different contexts and must be individually addressed during model development. Through controlled experiments on datasets of molecular properties, we show important trends in model performance associated with the level of noise in the dataset, size of the dataset, model architecture, molecule representation, ensemble size, and dataset splitting. In particular, we show that 1) noise in the test set can limit a model's observed performance when the actual performance is much better, 2) using size-extensive model aggregation structures is crucial for extensive property prediction, and 3) ensembling is a reliable tool for uncertainty quantification and improvement specifically for the contribution of model variance. We develop general guidelines on how to improve an underperforming model when falling into different uncertainty contexts.

  • Open Access

    Autonomous, multi-property-driven molecular discovery: from predictions to measurements and back

    by Brent Koscher, Richard B. Canty, Matthew A. McDonald, Kevin P. Greenman, Charles J. McGill, Camille L. Bilodeau, Wengong Jin, Haoyang Wu, Florence H. Vermeire, Brooke Jin, et al.

    A closed-loop, autonomous molecular discovery platform driven by integrated machine learning tools was developed to accelerate the design of molecules with desired properties. Two case studies are demonstrated on dye-like molecules, targeting absorption wavelength, lipophilicity, and photo-oxidative stability. In the first, the platform experimentally realized 312 unreported molecules across three automatic iterations of molecular design-make-test-analyze cycles while exploring the structure–function space of four rarely reported scaffolds. In each iteration, the property-prediction models which guided the exploration learned the structure–property space of diverse inexpensive scaffold derivatives realized through using multi-step syntheses. Conversely, the second study exploited property models trained on a chemical space with pre-existing examples to discover 6 top-performing molecules within the structure-property space. By closing the molecular discovery cycle of prediction, synthesis, measurement, and model retraining, the platform demonstrates the potential for integrated platforms to automatically understand a local chemical space and discover functional molecules.


  • DFT Analysis of Organotin Catalytic Mechanisms in Dehydration Esterification Reactions for Terephthalic Acid and 2,2,4,4-Tetramethyl-1,3-cyclobutanediol

    by Jennifer A. Clark, Pranav J. Thacker, Charles J. McGill, Jason R. Miles, Phillip R. Westmoreland, Kirill Efimenko, Jan Genzer, Erik E. Santiso

    Polyesters synthesized from 2,2,4,4-tetramethyl-1,3-cyclobutanediol (TMCD) and terephthalic acid (TPA) are improved alternatives to toxic polycarbonates based on bisphenol A. In this work, we use ωB97X-D/LANL2DZdp calculations, in the presence of a benzaldehyde polarizable continuum model solvent, to show that esterification of TMCD and TPA will reduce and subsequently dehydrate a dimethyl tin oxide catalyst, becoming ligands on the now four-coordinate complex. This reaction then proceeds most plausibly by an intramolecular acyl-transfer mechanism from the tin complex, aided by a coordinated proton donor such as hydronium. These findings are a key first step in understanding polyester synthesis and avoiding undesirable side reactions during production.