Molecule Generation of Drugs Using VAE
K. B. Anusha, Modalavalasa Divya, K. Madhuri Pravallikha Rani, B. Satvika, P. Tarun, G. Vaishnavi, S. Linga Raju · Advances in computer science research · 2024
The field of drug discovery and development has witnessed a transformative change during the previous few years with application in artificial intelligence and ML techniques.Among these, Variational Autoencoders (VAEs) have emerged promising instrument for the generative designing of drug molecules.In relation to drugs discovery, VAEs have been employed to encode and decode chemical structures, facilitating the generation of drug molecules.This is achieved through the encoding of chemical configurations into an uninterrupted latent space, where the generative capacity of the model can be harnessed to create diverse and potentially pharmacologically relevant compounds.Key components of this approach include the representation of molecules as graphs or SMILES (Simplified Molecular Input Line Entry System) strings, In development of specialized loss functions to optimize characteristics of molecules, and the investigating the latent space to produce molecules with desired characteristics.Hence, In this project, we build compounds for drug discovery using a variational autoencoder.