Explainable Deep Relational Networks for Predicting Compound-Protein Affinities and Contacts

Abstract

Predicting compound-protein affinity is critical for accelerating drug discovery.Recent progress made by machine learning focuses on accuracy but leaves much to be desired for interpretability. Through molecular contacts underlying affinities, our large-scale interpretability assessment finds commonly-used attention mechanisms inadequate. We thus formulate a hierarchical multi-objective learning problem whose predicted contacts form the basis for predicted affinities. We further design a physicsinspired deep relational network, DeepRelations, with intrinsically explainable architecture. Specifically, various atomic-level contacts or "relations" lead to molecular-level affinity prediction. And the embedded attentions are regularized with predicted structural contexts and supervised with partially available training contacts. DeepRelations shows superior interpretability to the state-of-the-art: without compromising affinity prediction, it boosts the AUPRC of contact prediction 9.5, 16.9, 19.3 and 5.7-fold for the test, compound-unique, protein-unique, and both-unique sets, respectively. Our study represents the first dedicated model development and systematic model assessment for interpretable machine learning of compound-protein affinity.

Knowledge-primed neural networks enable biologically interpretable deep learning on single-cell sequencing data

“…Pioneering research aimed at making deep learning models interpretable and informative for biological applications focused primarily on ex post analysis of trained ANNs, for example by identifying inputs that result in specific predictions [ 23 , 31 , 32 ] or by analyzing the compressed layers of autoencoders [ 33 ]. A complementary approach is the ex ante engineering of deep learning architectures for built-in biological interpretability, for example by including domain knowledge from structural biology [ 34 , 35 ], biophysical regulation of transcription [ 36 , 37 ], gene annotations [ 38 ], cancer growth [ 39 ], genetic screens [ 40 ], or by combining knowledge from different domains [ 41 – 43 ].…”

Cross-Modality Protein Embedding for Compound-Protein Affinity and Contact Prediction

“…We compare our single-modality and cross-modality models with two latest SOTAs for the CPAC problem, namely Gao et al [8] and DeepAffinity+ [12]. Tasks involved include affinity, contact, and binding-site predictions.…”

“…Ncomp×Nprot is targeted at making prediction for both the intermolecular affinity z aff and (atom-residue) contacts Z inter , where X comp , X prot are respectively the spaces for X comp , X prot . The SOTA pipelines for CPAC [12,9,13] comprise of the following three major components as shown in Figure 1.…”

“…After the CPAC model f CPAC forwardly generates the outputs (z aff , Z inter ), true labels (y aff , Y inter ) are compared to calculate the loss, l CPAC , which consists of affinity loss l aff , intermolecular atom-residue contact/interaction loss l inter and three structure-aware sparsity regularization loss l group , l fused , l L1 described in [12], expressed as:…”

“…Specifically, we aim at simultaneous prediction of compound-protein affinity and contacts in the aforementioned structurefree setting. We note that earlier works for this task represent proteins as 1D amino-acid sequences [12,13] or 1D structurally-annotated sequences [9]. However, 1D sequences of proteins adopt 3D structures to function, including interactions with compounds; so structure-aware representations of proteins (such as sequence-predicted residue-residue 2D contact maps) can also be useful, as explored in a recent affinity predictor [11].…”