AnoChem: Prediction of chemical structural abnormalities based on machine learning models

• DOI: 10.1016/j.csbj.2024.05.017
• Publication Date: 2024-05-16T01:03:28Z
• Authors: Changdai Gu; Woo Dae Jang; Kwang-Seok Oh; Jae Yong Ryu
• Keywords: 0301 basic medicine; Computational chemistry; 0303 health sciences; Cheminformatics; Machine learning; Method Article; AnoChem; Drug design; TP248.13-248.65; Biotechnology

drug design aims to rationally discover novel and potent compounds while reducing experimental costs during the development stage. Despite numerous generative models that have been developed, few successful cases of utilizing reported. One most common challenges is designing are not synthesizable or realistic. Therefore, methods capable accurately assessing chemical structures proposed by for needed. In this study, we present AnoChem, a computational framework based on deep learning designed assess likelihood generated molecule being real. AnoChem achieves an area under receiver operating characteristic curve score 0.900 distinguishing between real molecules. We utilized evaluate compare performances several models, using other metrics, namely SAscore Fréschet ChemNet distance (FCD). demonstrates strong correlation with these validating its effectiveness as reliable tool models. The source code available at https://github.com/CSB-L/AnoChem.