A Perspective on Explanations of Molecular Prediction Models

Wellawatte, Geemi P.; Gandhi, Heta A.; Seshadri, Aditi; White, Andrew D.

doi:10.1021/acs.jctc.2c01235

Cited by 27 publications

(31 citation statements)

References 139 publications

(344 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A range of XAI methods have been proposed in recent years , (Figure ) and are discussed in more detail in two recent comprehensive reviews. , Two simple strategies for XAI are to use self-explainable white-box methods and to check how changing inputs affect the outputs of black-box networks. Feature importance methods are a notable class of white-box methods that achieve explainability by identifying essential features based on model parameters such as weights and coefficients.…”

Section: Future Opportunitiesmentioning

confidence: 99%

Machine Learning-Guided Protein Engineering

Kouba,

Kohout,

Haddadi

et al. 2023

ACS Catal.

View full text Add to dashboard Cite

Recent progress in engineering highly promising biocatalysts has increasingly involved machine learning methods. These methods leverage existing experimental and simulation data to aid in the discovery and annotation of promising enzymes, as well as in suggesting beneficial mutations for improving known targets. The field of machine learning for protein engineering is gathering steam, driven by recent success stories and notable progress in other areas. It already encompasses ambitious tasks such as understanding and predicting protein structure and function, catalytic efficiency, enantioselectivity, protein dynamics, stability, solubility, aggregation, and more. Nonetheless, the field is still evolving, with many challenges to overcome and questions to address. In this Perspective, we provide an overview of ongoing trends in this domain, highlight recent case studies, and examine the current limitations of machine learning-based methods. We emphasize the crucial importance of thorough experimental validation of emerging models before their use for rational protein design. We present our opinions on the fundamental problems and outline the potential directions for future research.

show abstract

Section: Future Opportunitiesmentioning

confidence: 99%

Machine Learning-Guided Protein Engineering

Kouba,

Kohout,

Haddadi

et al. 2023

ACS Catal.

View full text Add to dashboard Cite

show abstract

“…This can be achieved using explainable artificial intelligence (XAI) methods, which provide a window into the ML model’s decision-making process and correlations uncovered by the model through data analysis . Justification and interpretability offered by XAI methods not only provide evidence defending why a prediction is trustworthy with quantitative metrics but also refer to the degree of human understanding intrinsic within the model. ,, …”

Section: Introductionmentioning

confidence: 99%

Integrating Explainability into Graph Neural Network Models for the Prediction of X-ray Absorption Spectra

Kotobi,

Singh,

Höche

et al. 2023

J. Am. Chem. Soc.

View full text Add to dashboard Cite

The use of sophisticated machine learning (ML) models, such as graph neural networks (GNNs), to predict complex molecular properties or all kinds of spectra has grown rapidly. However, ensuring the interpretability of these models' predictions remains a challenge. For example, a rigorous understanding of the predicted X-ray absorption spectrum (XAS) generated by such ML models requires an in-depth investigation of the respective black-box ML model used. Here, this is done for different GNNs based on a comprehensive, custom-generated XAS data set for small organic molecules. We show that a thorough analysis of the different ML models with respect to the local and global environments considered in each ML model is essential for the selection of an appropriate ML model that allows a robust XAS prediction. Moreover, we employ feature attribution to determine the respective contributions of various atoms in the molecules to the peaks observed in the XAS spectrum. By comparing this peak assignment to the core and virtual orbitals from the quantum chemical calculations underlying our data set, we demonstrate that it is possible to relate the atomic contributions via these orbitals to the XAS spectrum.

show abstract

“…In this way, we identify “Pareto-optimal” IDP sequences, meaning that sequence perturbations cannot enhance the dynamics further without reducing the stability of the condensate. Finally, we examine sequence features of Pareto-optimal sequences and perform a counterfactual analysis[35, 36] to identify the sequence determinants of the limiting thermodynamics–dynamics tradeoff. Taken together, our results demonstrate how sequence design can be used to tune thermodynamic and dynamical properties independently in the context of biomolecular condensates.…”

Section: Introductionmentioning

confidence: 99%

Active learning of the thermodynamics–dynamics tradeoff in protein condensates

An¹,

Webb²,

Jacobs³

2023

Preprint

View full text Add to dashboard Cite

Phase-separated biomolecular condensates exhibit a wide range of dynamical properties, which depend on the sequences of the constituent proteins and RNAs. However, it is unclear to what extent condensate dynamics can be tuned without also changing the thermodynamic properties that govern phase separation. Using coarse-grained simulations of intrinsically disordered proteins, we show that the dynamics and thermodynamics of homopolymer condensates are strongly correlated, with increased condensate stability being coincident with low mobilities and high viscosities. We then apply an "active learning" strategy to identify heteropolymer sequences that break this correlation. This data-driven approach and accompanying analysis reveal how heterogeneous amino-acid compositions and non-uniform sequence patterning map to a range of independently tunable dynamical and thermodynamic properties of biomolecular condensates.

show abstract

A Perspective on Explanations of Molecular Prediction Models

Cited by 27 publications

References 139 publications

Machine Learning-Guided Protein Engineering

Machine Learning-Guided Protein Engineering

Integrating Explainability into Graph Neural Network Models for the Prediction of X-ray Absorption Spectra

Active learning of the thermodynamics–dynamics tradeoff in protein condensates

Contact Info

Product

Resources

About