<br/>The rise of artificial intelligence in various scientific and technological areas can boost the drug discovery process. However, the potential of AI in medicinal chemistry has not been fully explored yet. Moreover, there is a lack of AI-based computational platforms that can provide the exploration of chemical space. In this talk, we present Syntelly: AI-inspired computational platform for organic and medicinal chemistry. Our platform provides a plethora of machine learning models for predicting physicochemical, biological (inhibition of CYP-proteins), and toxicological properties (developmental toxicity, acute toxicity on different animal species with various administration types) of organic compounds. Also, the platform provides standard tools that organic chemists need for everyday use, such as the generation of IUPAC names, the prediction of NMR spectra, and the estimation of synthetic complexity. For the navigation in chemical space. Syntelly provides a method for the projection of molecular sets to the 2D planes, grouping similar compounds together to visualize large chemical databases intuitively. Our platform is based on three pillars: clarity, modularity, and extensibility. To provide modularity, we use microservice architecture based on Docker and Singularity containers. Due to containerization, one can deploy the platform on HPC clusters or in clouds. We developed a python library “simply” to support the development of Syntelly plugins. At the talk, we demonstrate how one can write a plugin for Syntelly extending the platform's functionality.

Syntelly: AI-inspired computational platform for organic and medicinal chemistry

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.