An in silico Approach for Integrating Phenotypic and Target-based Approaches in Drug Discovery

Predicting molecular docking of per- and polyfluoroalkyl substances to blood protein using generative artificial intelligence algorithm DiffDock

This study computationally evaluates the molecular docking affinity of various perfluoroalkyl and polyfluoroalkyl substances (PFAs) towards blood proteins using a generative machine-learning algorithm, DiffDock, specialized in protein-ligand blind-docking learning and prediction. Concerns about the chemical pathways and accumulation of PFAs in the environment and eventually in the human body has been rising due to empirical findings that levels of PFAs in human blood has been rising. DiffDock may offer a fast approach in determining the fate and potential molecular pathways of PFAs in human body.

Coding recognition of the dose-effect interdependence of small biomolecules encrypted on paired chromatographic-based microassay arrays

The discovery of small biomolecules has suffered from the lack of a comprehensive framework to express the intrinsic correlation between bioactivity and the contribution from small molecules in complex samples with molecular and bioactivity diversity. Here, by mapping a sample’s 2D-HPTLC fingerprint to microplates, paired chromatographic-based microassay arrays are created, which can be used as quasi-chips to characterize multiple attributes of chromatographic components; as the array differential expression of the bioactivity and molecular attributes of irregular chromatographic spots for dose–effect interdependent encoding; and also as the automatic-collimated array mosaics of the multi-attributes of each component itself encrypted by its chromatographic fingerprint. Based on this homologous framework, we propose a correlating recognition strategy for small biomolecules through their self-consistent chromatographic behavior characteristics. In the approach, the small biomolecule recognition in diverse compounds is transformed into a constraint satisfaction problem, which is addressed through examining the dose–effect interdependence of the homologous 2D code pairs by an array matching algorithm, instead of preparing diverse compound monomers of complex test samples for identification item-by-item. Furthermore, considering the dose–effect interdependent 2D code pairs as links and the digital-specific quasimolecular ions as nodes, an extendable self-consistent framework that correlates mammalian cell phenotypic and target-based bioassays with small biomolecules is established. Therefore, the small molecule contributions and the correlations of bioactivities, as well as their pathways, can be comprehensively revealed, so as to improve the reliability and efficiency of screening. This strategy was successfully applied to galangal, and demonstrated the high-throughput digital preliminary screening of small biomolecules in a natural product.