Abstract: Drug induced gene expression provides information covering various aspects of drug discovery and development. Recent advances in accessibility of open-source drug-induced transcriptomic data along with ability of deep learning algorithms to understand hidden patterns have opened opportunity for designing drug molecules based on desired gene expression signatures. Embodiments herein provide method and system for cell specific model where gene expressions are processed via pretrained Simplified Molecular Input Line Entry System (SMILES) variational autoencoder (s-VAE) to produce new molecules. The model is trained with drug and drug induced gene expression data as input. Both pretrained s-VAE and profile variational autoencoder (p-VAE) are trained jointly. During joint training, difference between newly generated molecules and existing drug molecules is calculated as joint loss function composed of binary cross entropy loss and Kullback-Leibler divergence loss.

Abstract: Traditional drug discovery methods are target-based, time- and resource-intensive, and require a lot of resources for the initial hit molecule identification. Phenotype-based drug screening requires differential gene expression data of a large number of molecules for different combinations of cell-line, time point and dosage. Experimentally obtaining gene expression data for all these combinations is again a heavily resource-intensive process. The technical challenge in conventional methods that use prediction models is that they depend largely on the data processing and representation. The disclosure herein generally relates to drug-like molecule screening, and, more particularly, to a method and system for gene expression and machine learning-based drug screening. The embodiment, thus, provides a mechanism of a small molecule-induced gene expression prediction based on machine learning models.