DELvision: Improving Data-Driven Drug Discovery
DELvision offers high quality, off-the-shelf DEL screening dataset
WuXi AppTec, through its WuXi Biology business, is unveiling DELvision, a powerful DEL-based toolset for AIphafold-assisted drug discovery and target sciences.
Target-based drug discovery has been one of the most accessed venues in modern drug discovery. Successful target-based drug discovery begins with the identification of hit compounds that have the potential for good potency, selectivity and developability. It also requires understanding a target’s role in the disease to be addressed, as well as the target’s “druggability”. The development of large-scale screening and high-dimensional biology (so called “multi-omics”) tools enable more efficient target- and candidate-selection, but significant challenges remain.
DEL is an advanced screening technology that is widely used in hit-finding efforts. In contrast to high throughput screening, DEL requires a very small amount of target protein, and minimal assay development. More importantly, it disrupts the concept of “cost-per-well” and allows screening billions of compounds in one tube. With its huge compound library, each compound represents a small molecule probe. DEL is a remarkable approach to annotate small molecule and protein interaction, thus has the power of connecting the proteome with the vast small molecule chemical space.
The recent advancements in machine learning and virtual drug discovery are both promising and highlighting the need for more training data. To this end, WuXi Biology developed DELvision, an enhanced tool that integrates chemistry with target sciences. It comprises the small molecule binding profiles of 1500 (and growing) protein targets which have been subjected to interrogation in the same but vast chemical space of DEL libraries. In contrast to conventional screens where the target space is low-dimensional, DELvision comprises a high-dimensional map of binding relationships between small molecules and protein targets. As such, this rich data universe enables simultaneous characterization of targets and exploration of the target-modulating chemical space.
In addition to offering off-the-shelf, high quality DEL screening data covering thousands of targets, DELvision is positioned to provide valuable training data for machine learning of structure-activity relationships. The many-to-many binding relationships can reveal otherwise-hidden hit selectivity and promiscuity, allow refinement of binding pockets, and enable clustering of and differentiation within target families, among other applications. These functionalities complement and enhance structure prediction tools such as Alphafold by adding large volume of ligand binding information to facilitate modern CADD.
Recommended by LinkedIn
DELvision is built with optimized high-throughput protein production and DEL screening workflow
DELvision was made possible by taking advantage of WuXi Biology’s proprietary DEL and high-throughput protein production capabilities. The DELvision chemical libraries includes approximately 10,000 building blocks, covering various bi/tri-functional combinations and reactions. Furthermore, it comprises over 40 distinct compound libraries and roughly 14 billion compounds. In an effort to provide uniformity and consistency, all the targets that are used in DELvision are solely from internal sources, where the protein production and DEL screening were conducted for the purpose of DELvision platform only. More than 1,500 proteins, including kinases, E3 ligases, transcription factors, and the like, have been annotated. The number of covered protein and classes are increasing almost daily.
In order to produce protein in parallel, our scientists implemented an automated high-throughput protein production platform (HTPP), serving as the foundation of DELvision. The HTPP workflow streamlines the construct design, expression, separation, purification, and verification steps. The process is highly automated, starting from high efficiency AlphaFold-assisted construct design, followed by parallel production of hundreds of proteins at 100 ug scale. The protein produced through HTPP workflow is then QC-ed and sent to screening. By adapting to a high-throughput format, hundreds of proteins can be screened in parallel. The quality of DEL screening is also monitored closely to ensure data quality in the DELvision. Through a cohesive strategy encompassing protein sample production, parallel screening processes (such as protein dosage, buffer solution, immobilization strategy, and DEL library), and computational analysis workflows, DELvision generates readily accessible data that is of high quality and uniformity. This harmonized approach facilitates subsequent data processing, analysis, and utilization with utmost convenience. The whole process is also readily adaptive to custom library selection or other profiling assays.
At WuXi AppTec, doing the right thing for patients is behind everything we do. Through the unique CRDMO and CTDMO business models, our company is dedicated to lowering the barrier for new drug R&D and enabling partners to bring groundbreaking therapies to patients faster. WuXi AppTec will remain at the forefront of technological advancements, integrating platforms and enhancing diverse capabilities to better support the customers. Together, we can shape drug discovery and get closer towards a brighter future for all.