Introduction of CSTAR

RNAs play essential roles in biological processes and have emerged as promising targets for therapeutic interventions. Small molecules remain to be the major direction in drug discovery owing to their oral bioavailability and extensive knowledge in structure-activity relationship and medicinal chemistry for pharmacophore optimization with improved pharmacokinetics, specificity and potency. However, discovery of RNA-targeting small molecules remains challenging due to our limited understandings in the chemical space associated with specific RNA-binding. The current CSTAR database contains over 1.8 million Commercial Small molecules TArgeting RNA, which is curated by RiboBIND, a deep-learning transformer architecture utilizing 3D atom positions to extract RNA-binding features from small molecules. This webserver offers features including search, classification, download and prediction. Visitors can input a SMILES format file or draw a moiety structure to search within CSTAR compounds containing or similar to the input (sub)structure. The library or search results can be classified by vendor or properties such as molecular weight, logP values, hydrogen bond donors and acceptors. The resulting compound list can be downloaded in desired formats for future applications, such as virtual screening. Additionally, compound structures can be uploaded as input to predict their RNA-binding preferences.