Reduce drug discovery time to several months by applying AI-based natural language processing algorithms used in ChatGPT and Google search to amino acid sequences to create protein language models

Google search and chat AI 'ChatGPT' use natural language processing algorithms to mechanically process sentences written by humans. It has been reported that such natural language processing algorithms are also active in the drug discovery field.

How AI That Powers Chatbots and Search Queries Could Discover New Drugs - WSJ
https://www.wsj.com/articles/how-ai-that-powers-chatbots-and-search-queries-could-discover-new-drugs-11670428795

Medicines such as antipyretics and headache medicines contain proteins that affect various symptoms. method requires many years to derive the structure of the protein.

Proteins have a structure in which amino acids are arranged side by side, and deriving the protein structure is synonymous with deriving the amino acid sequence. Each amino acid can be abbreviated with one letter of the alphabet, such as A for alanine, R for arginine, and N for asparagine. For this reason, it is possible to apply a natural language processing algorithm by assuming the sequence of amino acids as a natural language. By training a natural language processing algorithm according to the sequence of amino acids, it becomes possible to derive amino acid sequences that can be used for new drugs in a few months by capturing sequences that are effective for specific symptoms as ``grammars''.

Already, the movement to utilize natural language processing algorithms in the field of drug discovery is progressing around the world, and expectations are high for the development of therapeutic drugs for diseases for which no effective drugs have been developed so far. Sean McLain, CEO of Absci , a pharmaceutical company based in Vancouver, Canada, said, ``The use of natural language processing algorithms will advance research in areas where no effective drugs existed before.'' .

As mentioned above, natural language processing algorithms are expected to speed up drug discovery. exists. For this reason, at the time of writing the article, research is being prioritized for applications such as ``fine-tuning the structure of existing drugs to improve their effectiveness''. For example, in a paper published in August 2022 by the aforementioned pharmaceutical company Absci, a natural language processing algorithm was used to adjust the structure of the anticancer drug trastuzumab to strengthen the binding between the protein and cancer cells. have succeeded in

Karen Hao, who reported on the use of natural language processing algorithms for drug discovery, said, ``The pharmaceutical industry is regulated by strict rules, and whether the products of natural language processing algorithms are effective or harmful. There are clear criteria by which we can judge, and I like this point.”

This is what I like about this particular application of large language models. The drug industry is highly regulated, and there are clear ways to measure whether the output of a protein-language model is doing more harm than good.

— Karen Hao 郝珂灵 @[email protected] (@_KarenHao) December 7, 2022