canSAR 2024-an update to the public drug discovery knowledgebase

Abstract

canSAR (https://cansar.ai) continues to serve as the largest publicly available platform for cancer-focused drug discovery and translational research. It integrates multidisciplinary data from disparate and otherwise siloed public data sources as well as data curated uniquely for canSAR. In addition, canSAR deploys a suite of curation and standardization tools together with AI algorithms to generate new knowledge from these integrated data to inform hypothesis generation. Here we report the latest updates to canSAR. As well as increasing available data, we provide enhancements to our algorithms to improve the offering to the user. Notably, our enhancements include a revised ligandability classifier leveraging Positive Unlabeled Learning that finds twice as many ligandable opportunities across the pocketome, and our revised chemical standardization pipeline and hierarchy better enables the aggregation of structurally related molecular records.

Graphical Abstract

Figure 1.

Structural ligandability for KRAS highlights accessibility to ligandable structures and pockets. (A) Color coding of ligandability and structural alignment by sequence and functional domains enables discovery of structures of interest and shows that there are 656 experimentally determined structural instances for KRAS, of which 535 are predicted to have a ligandable cavity (green). The top bar for every protein represents the AlphaFold2 model. (B) Each individual structure can be examined in detail. Ligandable pockets are visualized with present ligands and physicochemical property distributions.

Figure 2.

The chemical hierarchy for pacritinib illustrates how related structural forms of compounds as well as their associated bioactivities are aggregated. (A) Hierarchical ordering of molecules affords traceability of chemical records according to the performed chemical standardization transformations. (B) Bioactivities are integrated alongside associated chemical structural data, allowing for discovery of related records and activities.

Figure 3.

Scaffold searching for sorafenib tosylate. Fragments and isotopic labeling are now correctly removed from Bemis-Murcko Scaffolds. This allows for improved discovery of structurally related compounds and, by extension, associated bioactivities.

Figure 4.

The ‘Mutations’ page for KRAS. This visualization integrates mutational frequency and types, functional annotations, protein–protein interaction sites and ligandability predictions, allowing users to identify opportunities in sequence space with likely functional consequence.