How Artificial Intelligence can speed up drug discovery

April 08, 2020 Lokesh kumar 0 Comments


It usually takes a drug around ten years to reach the market since its initial discovery. The initial clinical trials take six to seven years on an average. The average cost of drug discovery is around $2.6 billion. These figures leave nothing for me to explain regarding the difficulties of drug discovery. However, the situation is changing and we are entering an era when these hardships could be history.


A Hongkong based pharmaceutical company called Insilico Medicine worked with the researchers from Toronto university to design 30,000 designs for molecules that can challenge a protein related to fibrosis in September of 2019. It took them 21 days. Then they isolated six molecular structures and then picked two from those. The most potent molecule was tested on a rat with favorable results. The most exciting part of this story is that the whole process took no more than 46 days. Yes, we are talking about the conduction of six to seven years worth of research and development in 46 days. This miraculous feat was realized with the help of an AI driven system called Generative Tensorial Reinforcement Learning or GTRL.

Integrating AI in the drug discovery process


The kind of research that was referred to in the last section is still not second nature to the pharmaceutical industry. Things in this industry are more complicated than just inventing a new tool and solving the problem forever. Nevertheless, innovative steps are being taken quite consistently. Various technologies used in the process of drug design are being infused with artificial intelligence.

Augmenting the information engine


The information engine is a system used by researchers. This gathers and stores data from different sources, data that is relevant to the efforts of drug discovery and pharmaceutical research. The data that is collected includes everything from research works done in different universities to patents owned by other big companies. The information aggregated by the information engine can be sorted and processed faster with the help of AI.

Speeding up research 


AI based systems can work directly with the molecular structures of drug candidates. Finding the appropriate target (that is the disease which can be cured with the drug), locating pathways for repurpusing (using an already existing compound for a different outcome) an existing drug, all of these can be brought to speed.

Fighting the Corona crisis with AI


Google deepmind has already used its AI systems to identify six protein structures that may have constructed the capsule outside the COVID-19 virus. Although this information is hypothetical still it can be crucial in finding a cure. Insilico Medicine has found molecular structures that can break the protein on the COVID-19. Germany based startup Innoplexus has developed a molecule that has binding quality against a protein on the virus. Other startups are trying to find ways of repurpusing already existing drugs to fight Corona. This is a real test for the AI based drug discovery startups, which have been heavily funded in recent times and employ more than 10,000 personnel world wide. It is also an awakening for the AI enthusiasts to undergo an applied artificial intelligence course and join the cohorts of this burgeoning industry. 

 


0 comments: