Revolutionising drug discovery with cutting-edge technologies
Researchers are empowering drug discovery with artificial intelligence (AI) and bioinformatics tools, unlocking new frontiers in drug design. Using AlphaFold, PandaOmics and Chemistry42, scientists have found new ways to predict protein shapes and design medicines for conditions typically challenging to treat. This ground-breaking research could be instrumental in combatting diseases, particularly cancers.
Chemical scientists have used computer systems and tools to design drugs since the 2000s. Until now, there have been several hurdles. Many systems still have been unable to accurately predict the behaviour of medicines.
Emerging technologies including AI have been pivotal in advancing drug discovery efforts. In 2020, a computer program named AlphaFold incredibly predicted the shape of proteins in the whole human genome, some of which had never been seen before. This revolutionary discovery had major implications for the future of treating diseases, allowing scientists to design new medicines faster than ever before.
Unleashing the power of AI to combat disease
Feng Ren and co-workers used AlphaFold alongside two other key platforms in their research. PandaOmics functions as a database of protein shapes, allowing scientists to compare proteins from AlphaFold to those already in PandaOmics. Through this, they found proteins only present in cancer cells. Finally, they also used Chemistry42, a platform that utilises AI to optimise and design small molecules. With the help of Chemistry42, they could apply the protein shapes and then design medicines that combat the proteins found.
After analysing text and data on total ribonucleic acid (RNA) and protein dynamics (OMICs) from 10 datasets for the human cancer hepatocarcinoma (HCC), PandaOmics provided a list of the top 20 targets with unknown protein folds.
AlphaFold was used to predict the structures of the target proteins. The team focused on the regulatory protein cyclin-dependent kinase 20 (CDK20), a protein-coding gene associated with several diseases with few associated approved drugs in the last three years, as a target for this research.
Pioneering new approaches to drug design
Using Chemistry42 to generate compounds based on the predicted CDK20 structure from AlphaFold, more than 8918 molecules were generated. One compound, ISM042-2-001 was particularly successful in cancer cell growth inhibition.
Further experimentation following this discovery enabled the team to generate further compounds, eventually finding the molecule ISM042-2-048, a molecule that performed 15 times better than the first molecule discovered.
... this work represents the first example of successfully utilising AlphaFold predicted protein structures for hit identification for a novel target. Further applications of this approach to other target classes such as GPCR and E3 ligase are ongoing.
This research could be revolutionary in developing life-saving drugs for a wide variety of conditions. By incorporating the practices outlined in this research further, scientists may be able to generate further medicines to treat a wider variety of diseases.
This article is free to read in our open access, flagship journal Chemical 新月直播app下载: Kim et al, Chem. Sci., 2023.