MolTrans: Molecular Interaction Transformer for drug–target interaction prediction

Abstract

Motivation Drug target interaction (DTI) prediction is a foundational task for in-silico drug discovery, which is costly and time-consuming due to the need of experimental search over large drug compound space. Recent years have witnessed promising progress for deep learning in DTI predictions. However, the following challenges are still open: (1) existing molecular representation learning approaches ignore the sub-structural nature of DTI, thus produce results that are less accurate and difficult to explain; (2) existing methods focus on limited labeled data while ignoring the value of massive unlabelled molecular data. Results We propose a Molecular Interaction Transformer (MolTrans) to address these limitations via: (1) knowledge inspired sub-structural pattern mining algorithm and interaction modeling module for more accurate and interpretable DTI prediction; (2) an augmented transformer encoder to better extract and capture the semantic relations among substructures extracted from massive unlabeled biomedical data. We evaluate MolTrans on real world data and show it improved DTI prediction performance compared to state-of-the-art baselines. Availability The model scripts is available at https://github.com/kexinhuang12345/moltrans. Contact jimeng@illinois.edu Supplementary information Supplementary data are available at Bioinformatics online.

DTI-Voodoo: machine learning over interaction networks and ontology-based background knowledge predicts drug–target interactions

“…One major component of our experiments was to determine how the information that is available to a machine learning model during training affects the performance of the model. Similarly to previous work (Huang et al ., 2020; Lee et al ., 2019), we find significant differences in predictive performance across different splitting schemes.…”

“…Several of the biases we identify in evaluating DTI prediction methods have been observed previously. The performance difference based on how training and evaluation data is split (by interaction pair, by drug, or by protein) has been demonstrated before using the MacroAUC measure (Huang et al, 2020;Pahikkala et al, 2014;van Laarhoven and Marchiori, 2014); we further extend on these results by introducing performance measures based on micro averages (Micro AUCp and Micro AUC d ) to further illustrate how prediction performance changes when evaluation data is imbalanced. We have further extended on prior results by introducing a "naïve" classifier that explicitly exploits one data bias to make predictions, illustrating that this bias has a significant impact on DTI methods.…”

X-DPI: A structure-aware multi-modal deep learning model for drug-protein interactions prediction

“…5. MolTrans [9] is an end-to-end biological-inspired deep learningbased framework that models the DPI process. We followed the same hyper-parameter setting described in the paper and compare our model with the MolTrans on our dataset.…”

“…In this case, molecules can be characterized by molecular fingerprints, structural descriptors, or topographies, while proteins can be described by sequences or tertiary structures. Their representations are then extracted by the designed neural network to obtain abstract information and are eventually used to predict whether and how they will bind to each other [5,[8][9][10][11][12].…”

Prediction of Drug–Target Interactions by Combining Dual-Tree Complex Wavelet Transform with Ensemble Learning Method

“…Ezzat et al [ 21 ] presented a framework that combined feature dimensionality reduction and the ensemble learning model for predicting DTIs. Huang et al [ 22 ] developed a method called MolTrans (Molecular Interaction Transformer) to predict DTIs that combined the interaction modeling module and sub-structural pattern mining algorithm. Zhang et al [ 23 ] developed a method called SPVes that combined SMILES2Vec and ProtVec to convert SMILES strings of drug compounds and sequences of target proteins as feature vectors to predict DTIs.…”