Parallelised DNA encoded library screens with Artificial Intelligence

利用人工智能进行并行 DNA 编码文库筛选

• 批准号: 10068959
• 金额: $ 34.16万
• 依托单位: DEEPMIRROR LTD
• 依托单位国家: 英国
• 项目类别: Collaborative R&D
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=10068959
• 关键词: Parallelised; DNA; encoded; library; screens

Protein kinases are involved in almost all bodily function and changes in their activity can give rise to cancer and other diseases. Nevertheless, few drugs exist that target kinases as they are considered "difficult to drug" by leading experts.Recently, DNA Encoded Small Molecule Libraries (DELs) have attracted attention due to their ability to enable screening of trillions (~1,000,000,000,000) of potential drugs in a single experiment. While they would be exceptionally suited for discovering new drugs for hard-to-drug targets such as kinases, using them successfully remains difficult: DEL screens typically generate many potential drug candidates, but there is no easy way of developing one of them into a drug than is safe enough for human treatment.DeepMirror and Nuclera are developing disruptive new technologies to address these challenges. DeepMirror's software platform automates the development of machine learning models for drug discovery and suggests optimal lead candidates based on laboratory data such as DEL screens. Nuclera has developed the eProteinDiscoveryTM rapid protein prototyping system: an easy-to-use, pipette and forget, benchtop system that provides medium throughput numbers of proteins at microgram to milligram quantities for drug discovery applications. Combining these two technologies would enable a synergistic loop with which one could screen many different kinases in parallel, and then build AI models to select small molecules with low predicted cross-reactivity to test in the next screen, thus rapidly converging to the most promising small molecule leads.We are applying for the Biomedical Catalyst Grant to perform the essential R&D steps to upgrade and integrate our respective platforms for DEL screening execution and analysis. Our teams have deep and unique expertise in AI, protein synthesis, and laboratory experiments and we are supported by investors and advisors that hold senior positions in the pharma sector. The total taxpayer cost is estimated at £270k over the course of 23 months. If successful, our technologies will disrupt the early drug discovery phases for difficult-to-drug proteins, dramatically accelerating the development of new medicines against debilitating diseases such as cancer, with huge benefits for human health and the UK economy.

蛋白激酶几乎参与了所有的身体功能，其活动的变化会引起癌症和其他疾病。然而，很少有药物是针对激酶的，因为它们被主要专家认为是“难以药物”。近年来，DNA编码小分子文库（DNA Encoded Small Molecule Libraries, DELs）由于能够在单次实验中筛选数万亿（~1,000,000,000,000）种潜在药物而引起了人们的关注。虽然它们将特别适合于发现针对激酶等难以药物靶点的新药，但成功使用它们仍然很困难：DEL筛选通常会产生许多潜在的候选药物，但没有一种简单的方法可以将其中一种药物开发成足够安全的药物用于人类治疗。DeepMirror和Nuclera正在开发颠覆性的新技术来应对这些挑战。DeepMirror的软件平台可以自动开发用于药物发现的机器学习模型，并根据实验室数据（如DEL筛选）推荐最佳候选药物。Nuclera开发了eProteinDiscoveryTM快速蛋白质原型系统：一种易于使用的移液和遗忘式台式系统，可为药物发现应用提供微克至毫克量的中等通量蛋白质。将这两种技术结合起来，将形成一个协同循环，人们可以同时筛选许多不同的激酶，然后建立人工智能模型，选择交叉反应性较低的小分子，在下一个筛选中进行测试，从而迅速收敛到最有希望的小分子先导物。我们正在申请生物医学催化剂拨款，以执行必要的研发步骤，升级和整合我们各自的DEL筛选执行和分析平台。我们的团队在人工智能、蛋白质合成和实验室实验方面拥有深厚而独特的专业知识，我们得到了在制药行业担任高级职位的投资者和顾问的支持。在23个月的时间里，纳税人的总花费估计为27万英镑。如果取得成功，我们的技术将颠覆难以药物化的蛋白质的早期药物发现阶段，极大地加速针对癌症等使人衰弱的疾病的新药的开发，为人类健康和英国经济带来巨大利益。