Deep Learning-Based Ligand Design Using Shared Latent Implicit Fingerprints from Collaborative Filtering

In their previous work, Srinivas et al. [ J. Cheminf. 2018, 10, 56] have shown that implicit fingerprints capture ligands and proteins in a shared latent space, typically for the purposes of virtual screening with collaborative filtering models applied on known bioactivity data. this we extend these fingerprints/descriptors using deep learning techniques to translate descriptors into discrete representations molecules (SMILES), without explicitly optimizing chemical properties. This allows design new compounds based upon representation nearby proteins, thereby encoding druglike properties including binding affinities proteins. The descriptor method does not require any fingerprint similarity search, which makes free bias arising from empirical nature [Srinivas, R.; 56]. We evaluate potentially novel drugs generated by our approach physical complexity. Additionally, analyze reliability biological activity employing protein–ligand interaction, assists assessing potential affinity designed compounds. find exhibit chemically feasible are predicted be excellent binders Furthermore, also diversity created Tanimoto distance conclude there is wide