Predicting Critical Properties and Acentric Factors of Fluids Using Multitask Machine Learning

Abstract

Knowledge of critical properties, such as critical temperature, pressure, density, as well as acentric factor, is essential to calculate thermo-physical properties of chemical compounds. Experiments to determine critical properties and acentric factors are expensive and time intensive; therefore, we developed a machine learning (ML) model that can predict these molecular properties given the SMILES representation of a chemical species. We explored directed message passing neural network (D-MPNN) and graph attention network as ML architecture choices. Additionally, we investigated featurization with additional atomic and molecular features, multitask training, and pretraining using estimated data to optimize model performance. Our final model utilizes a D-MPNN layer to learn the molecular representation and is supplemented by Abraham parameters. A multitask training scheme was used to train a single model to predict all the critical properties and acentric factors along with boiling point, melting point, enthalpy of vaporization, and enthalpy of fusion. The model was evaluated on both random and scaffold splits where it shows state-of-the-art accuracies. The extensive data set of critical properties and acentric factors contains 1144 chemical compounds and is made available in the public domain together with the source code that can be used for further exploration.

ConfSolv: Prediction of Solute Conformer-Free Energies across a Range of Solvents

“…Others , have suggested the use of advanced aggregation techniques, such as attention mechanisms, to mix feature channels between solvent and solute, but our initial experiments suggested concatenation to be more robust. Vermeire and Green found that using additional features such as polar surface as computed by RDKit can improve performance, and several other studies have also explored augmenting D-MPNNs with a variety of heuristic, quantum mechanical, and chemical descriptors. − However, the outcome of using descriptors on model performance is not very consistent, and given the large number of solute conformers in this work, we do not use additional features in this work. We do note that the aforementioned studies are for 2D D-MPNNs and that exploring how additional descriptors impact the model architecture adopted here could be a topic of future work.…”

New functional form to describe the temperature dependence of liquid phase reaction rates

“…With a set of four equations (eq 12 -15) that are linear with respect to 𝑎, 𝑏, 𝑐, and 𝑑, we can solve for those four parameters based on ΔΔ𝐺 ‡ solv (298 K), ΔΔ𝐻 ‡ solv (298 K), and the saturation densities (𝜌 l (𝑇 )) and critical properties (𝑇 c , 𝜌 c ) of a solvent. The densities and properties of many common solvents can be obtained using open-source models, such as fluid thermodynamics packages CoolProp [33] and Clapeyron.jl [34], and a machine learning model developed by Biswas et al [35] Since the solvent's properties are easily obtainable from the existing models, we can treat them as known values. Thus, only ΔΔ𝐺 ‡ solv (298 K) and ΔΔ𝐻 ‡ solv (298 K) are needed to solve for the four empirical parameters and hence obtain ΔΔ𝐺 ‡ solv at any temperature.…”

Comment on ‘Physics-based representations for machine learning properties of chemical reactions’

“…If we are interested in assessing model performance on new molecules, we can train a model with many reaction templates but use substructure splitting to create training, validation, and testing sets. Bemis-Murcko scaffolds [70] are commonly used to partition the data for this purpose, though clustering based on other input features or chemical similarity to measure extrapolation has also been explored [23,[71][72][73][74][75][76][77][78][79][80][81][82][83][84][85][86][87][88] as has quantifying domains of model applicability [89][90][91][92][93]. Scaffold splitting is not perfect, but by ensuring that molecules in the testing set are structurally different than those in the training set, it offers a much better assessment of generalizability than splitting randomly [17,24,67,[94][95][96][97][98][99][100][101][102][103][104][105][106][107][108][109]…”