Accelerating De Novo Drug Design against Novel Proteins Using Deep Learning

Abstract

In the world plagued by the emergence of new diseases, it is essential that we accelerate the drug design process to develop new therapeutics against them. In recent years, deep learning-based methods have shown some success in ligand-based drug design. Yet, these methods face the problem of data scarcity while designing drugs against a novel target. In this work, the potential of deep learning and molecular modeling approaches was leveraged to develop a drug design pipeline, which can be useful for cases where there is limited or no availability of target-specific ligand datasets. Inhibitors of the homologues of the target protein were screened at the active site of the target protein to create an initial target-specific dataset. Transfer learning was used to learn the features of the target-specific dataset. A deep predictive model was utilized to predict the docking scores of newly designed molecules. Both these models were combined using reinforcement learning to design new chemical entities with an optimized docking score. The pipeline was validated by designing inhibitors against the human JAK2 protein, where none of the existing JAK2 inhibitors were used for training. The ability of the method to reproduce existing molecules from the validation dataset and design molecules with better binding energy demonstrates the potential of the proposed approach.

OdoriFy: A conglomerate of artificial intelligence–driven prediction engines for olfactory decoding

“…OR Finder prediction engine harbors functionalities to link the extranasal ORs to that of user-provided metabolites ( 73 ). Last and most importantly, the underlying deep learning frameworks of OdoriFy could be translated into other subdomains of chemoinformatics research such as in silico drug design ( 74 ).…”

Using Domain-Knowledge to Assist Lead Discovery in Early-Stage Drug Design

“…-Following [9], θ = 6.0. That is, all molecules with pIC50 value ≥ 6.0 are taken as "active" inhibitors; -The discriminator in Step 2c is a BotGNN.…”

“…There are assessment is done by one of the authors (AR), who is a computational chemist. The assessment uses structural features and functional groups identified for the JAK2 site in the literature [9,19,20]. also attempts to build molecule generation models against novel target proteins, where there is a limited ligand dataset for training the model [9,26].…”

“…Some works focused towards drug-like property optimization which helped in biasing the models to generate molecules with the properties of interest [22,24,25]. The efficiency of the models to generate chemically valid molecules with optimized properties has significantly improved [9,24]. There are 6 A good reason to consider dissimilar molecules is that it allows us to explore more diverse molecules.…”

Lib-INVENT: Reaction Based Generative Scaffold Decoration for in silico Library Design

“…

The ChEMBL databased is pre-processed using standard approaches adopted by many authors, such as Krishnan et al 1 . The aim of this procedure is to remove any compounds which could potentially direct the training of the prior model away from the desirable, drug-like chemical space which it is meant to learn.

…”