The Role of Tech and AI in Drug Discovery Today

Dr. Venkat engaged in a conversation with the India Pharma Outlook magazine in order to answer queries on how AI technologies and innovations are being used in drug discovery processes. He is serial entrepreneur and industry innovator. Dr. Venkat has proven history of driving breakthrough innovations and industry-first semiconductor products, first in wireless communications and now applying leadership expertise to redefine the future of AI and machine learning with an ultra-low energy supercomputing chip. He is the author of more than 100 patents with the vision, leadership and expertise capable of building companies from the grounds up and into multi-million global powerhouses that disrupt industries.

Can the AI systems for acceleration and stimulation in the drug discovery process provide support for reverse engineering processes or only for new drug discoveries?

The technology is applicable for both. A computing system can be used for the purpose of doing any specific function well and at the same time for hacking function as well. It is because computing resource doesn’t differentiate what function you are trying to perform. A computer will not have information about what the user is trying to do or whether it is reverse or forward engineering. Everything depends on the source. Both new drug discovery and reverse engineering require huge amount of data storage and computing requirements. The need of AI to reduce the problem size is paramount. The AI based supercomputing technologies are able to make the development process cost effective and at the same time less time consuming for developing a new drug.

What are some of the key challenges that AI helps to address in drug discovery?

The biggest challenge in drug discovery today is the time taken. The number of new drugs entering the market compared to the number of drugs required is very less. Moreover, for the phase 2 clinical trials, the failure rate is over 90% percent which is prohibitively high. The possibilities of creating options in the drug discovery space for a particular kind of drug are huge. If one takes the example of COVID-19, the possible proteins that can be created from the amino acid sequences can be in the order of 10 powers 600.  The filtering to come up with an optimal choice is not feasible with any computing resource available today. This leads to undesired side effects.  A major problem that needs to be addressed here is bringing down the complexity of the size of the problem and that is where AI comes into the picture to reduce the problem size and complexities.

What role do big data and machine learning play in analyzing vast amounts of biological and chemical data for drug discovery?

Megawatts of power, millions of dollars and months of time were generally required for a single training of the ChatGPT type of training. That is minuscule compared to the computing demand that is required for AI in order to emulate human cell for structure prediction. As I have mentioned that there are mainly two phases in a drug discovery process. First is the discovery phase and the next one is the clinical trials phase. Clinical trial involves huge amount data analysis and compilation and the amount of money spent is also a lot. In today’s methodology, billions of dollars are spent for one drug FDA approval and commercialization. AI can assist in speed up of the development time and it is proved in the COVID vaccine development. On the other hand, if data analytics and compilation of data is done manually, it takes a lot of time. AI systems are able to process these operations faster and are also able to detect errors in the analytics and solve them during the process itself. So, AI is able to perform analytics faster and can also prove to be more intelligent.  The data and the analytics provided by AI systems are more accurate. To be able to use AI in discovery phase depends on what drug we are developing and what accuracy we need. There is a huge progress and recent development in this space are very significant. 

Are there any notable success stories or breakthroughs achieved through the application of AI in drug discovery?

One of major challenges in drug discovery is the time taken. If an experiment takes a weeks’ time manually, the discovery and development process will also be delayed. Here, if a tool is provided which is able to perform the stimulation much faster and compile and analyze years of data, it means that this will help in development of better drugs with less amount of toxicity. AI stimulation systems that have emerged in the market are able to execute processes a 100 times faster. So, if an experiment takes 100 days manually, these systems are able to accomplish the same in only one day. It is true that all the problems regarding drug discovery and development has not yet been solved by AI but many tools and methodologies have come into the picture that can help find solutions to all the problems. The industry is currently developing more infrastructures in this regard. We are doing our part to accelerate simulation and emulate the entire system with hardware, software and AI technology to create an infrastructure for rapid drug development for the future.