Perceiver CPI: a nested cross-attention network for compound–protein interaction prediction

Abstract

Motivation Compound-protein interaction (CPI) plays an essential role in drug discovery and is performed via expensive molecular docking simulations. Many artificial intelligence-based approaches have been proposed in this regard. Recently, two types of models have accomplished promising results in exploiting molecular information: graph convolutional neural networks that construct a learned molecular representation from a graph structure (atoms and bonds), and neural networks that can be applied to compute on descriptors or fingerprints of molecules. However, the superiority of one method over the other is yet to be determined. Modern studies have endeavored to aggregate information that is extracted from compounds and proteins to form the CPI task. Nonetheless, these approaches have used a simple concatenation to combine them, which cannot fully capture the interaction between such information. Results We propose the Perceiver CPI network, which adopts a cross-attention mechanism to improve the learning ability of the representation of drug and target interactions and exploits the rich information obtained from extended-connectivity fingerprints to improve the performance. We evaluated Perceiver CPI on three main datasets, Davis, KIBA, and Metz, to compare the performance of our proposed model with that of state-of-the-art methods. The proposed method achieved satisfactory performance and exhibited significant improvements over previous approaches in all experiments. Availability Perceiver CPI is available at https://github.com/dmis-lab/PerceiverCPI Supplementary information Supplementary data are available at Bioinformatics online.

ChemGLaM: Chemical-Genomics Language Models for Compound-Protein Interaction Prediction

“…PerceiverCPI [32] is a state of the art model based on the nested cross-attention network that combines the representation of 2D graphs, molecule fingerprints, and protein sequences.…”

Drug-Target Interactions Prediction at Scale: the Komet Algorithm with the LCIdb Dataset

“…The method that is used to define this latent space has a critical impact on the prediction performance, and a key aspect is that the features representing the (molecule, protein) pair should capture information about the interaction, which is not fully achieved by simple concatenation between molecule and protein features. 25 Therefore, step 2 usually consists of a non-linear mixing of the protein and molecule embeddings, to better encode information about interaction determinants. One common approach is to use the tensor product, which is equivalent to a Kronecker kernel.…”

SadNet: a novel multimodal fusion network for protein–ligand binding affinity prediction

“…Perceiver CPI and GEFA also put forward a similar point of view: the fusion ability of concatenation is not enough. 25,26 From the perspective of model performance, compared with Pafnucy, the accuracy of DLSSAffinity in RMSE is only improved by 1.4%, which is not enough to illustrate the necessity of complementarity between three-dimensional atomic information and sequence information.…”