Beyond Human Limits: How AI is Accelerating Scientific Progress

Dr Ashok Kumar, President-CRD at IPCA Laboratories

The tide of the technological revolution is rising, with Artificial Intelligence (AI) at the forefront. Unlike past innovations that attempted to replace conventional tools entirely, AI acts as a collaborator. AI is an intelligent partner that excels at data analysis and continuous learning, fundamentally reshaping industries. From self-driving cars to the language prowess of ChatGPT-4, AI's impact is undeniable. It is a  powerful tool that not only answers questions but also creates content, solves complex problems, and even assists with medical assessments. As AI continues its rapid evolution, the key lies in responsible integration to unlock a future brimming with human-AI collaboration and groundbreaking discoveries.

Enhancing Scientific Research 

Traditionally, people were under the impression that creative and innovative vocations, including the fine arts, science, and complicated decision-making requiring critical thinking, are exclusively human domains and will never be impacted by AI. But, in the present scenario, this theory seems to be incorrect, as exemplified by a few recent developments in chemistry and related fields.

Discovering New Drugs

Unlocking new drugs hinges on understanding protein structures, a historically difficult task. Here, AI is making waves in this field as well. AlphaFold, a powerful algorithm, predicts protein structures from amino acid sequences in mere hours, slashing years off the traditional method. This breakthrough opens doors for a surge in new drug targets. AlphaFold's impact extends further. When combined with other AI tools like PandOmics and Chemistry 42, it can pinpoint potential drug candidates by analysing their predicted protein structures. The synergy between AI and humans is paving the way for a revolution in drug discovery, with the potential to expedite the development of life-saving medications.

"Drug discovery is on the cusp of a new era. By combining human expertise with AI's capabilities, we can accelerate the development of life-saving medications."            

New Drug Development

Along with drug discovery, AI is also revolutionizing drug development. Beyond early-stage target identification, AI tackles lead optimization, drug property prediction, and even clinical trial design. The combination of AI and Big Data cuts costs, reduces late-stage failures, and streamlines the entire process. Bringing a drug to the market takes over a decade and billions of dollars. However, with AI's help, the future of drug development is much faster and more efficient. Industry reports predict that within ten years, AI will be instrumental in the development of over half of all approved medications.

Predicting Enzyme Function

A new free online tool called CLEAN (contrastive learning enabled enzyme annotation) uses deep learning to analyse enzyme amino acid sequences and predict their function. It helps researchers quickly identify the right enzyme for specific chemical production.

A study reports that CLEAN outperforms previous EC number annotation methods in challenges including BLASTp and state-of-the-art ML models. 

Enzyme Engineering

AI now plays a crucial role in tackling the issue of plastic pollution. An AI-designed enzyme can break down entire plastic trays in under 48 hours, significantly faster and at lower temperatures than existing methods. This innovation outperforms human-made solutions for 51 different PET products, offering a promising new approach to plastic waste management. 

De Novo Enzyme Design

Proteins have complex 3D structures, making them difficult to design. However, AI is changing the game. Researchers can now use AI platforms like ColabFold to design protein sequences and predict their structure, bypassing limitations in our understanding of protein physics. It opens doors for creating novel proteins with specific functions, like the curative proteins designed by the AI model CHROMA.

Synthetic Route Design

Tools like Synthia (Merck KGAA) leverage AI and vast databases to recommend reaction pathways for complex molecules. It goes beyond existing literature, suggesting novel strategies for faster and more efficient synthesis. A 2023 study reported that AI-powered tools like AstraZeneca's Synthia platform can achieve success rates of up to 80% in predicting efficient synthetic routes for complex molecules, compared to traditional methods with a success rate closer to 50%. 

Similarly, IBM's ROBOT RXN predicts millions of potential reactions, accelerating the discovery of crucial chemical compounds used in drug development. These AI-powered tools are empowering chemists worldwide, with over 30,000 already utilising ROBOT RXN to streamline their work.

Reaction Condition Predictor

Human chemists are at risk of getting outperformed by AI chemists. AI robots predict optimal reaction conditions and perform experiments, achieving double the yield in Suzuki-Miyamura couplings. 

                     

For other reactions, AI can analyse vast literature (e.g., 62,000+ Buchwald-Hartwig reactions) to suggest reaction tweaks and prevent unwanted side effects. Labmate, another AI tool, recommends reaction conditions and even suggests further experiments to maximise yield.

                       

 

Enantioselectivity Prediction

AI can be trained to determine the activation energy for the competing R and S pathways, which can then be translated to enantiomeric excess, providing remarkable results in predicting the enantioselectivity of an asymmetric transformation.

Atom Tracking

AI chemists are making waves in scientific research. RXN Mapper tracks reaction atoms with over 99% accuracy, and an AI robot chemist discovered a photocatalyst six times better than human-made versions.

Stereospecific C-C Bond Formation

Scientists have created a fully automated robotic platform that can precisely build the core component of kalkitoxin, a neurotoxin naturally produced by cyanobacteria. This innovative system allows for the controlled growth of carbon chains one atom at a time, with the ability to manipulate both chain length and stereochemistry.

            

 

Self-driving Chemistry Labs

An AI chemist, Coscientist not only plans experiments, writes codes and analyzes results but also performs complex reactions like Suzuki and Sonogashira couplings without human input.  Similarly, ChemCrow, an advanced chatbot, can design syntheses and operate robotic labs. Given the prompt to create an insect repellent, ChemCrow autonomously produced DEET, a known repellent. These AI advancements are revolutionizing chemical research.

                                          

AI chemists like AlphaFlow are accelerating material discovery. It explores over a trillion potential quantum dot combinations in a month, autonomously optimising reaction conditions. Innovations such as these pave the way for the faster development of new materials.

Conclusion

AI is revolutionising scientific research. Its incredible learning speed, data analysis, and ability to tackle complex problems unlock new insights from experiments and vast datasets. The future promises even more with self-thinking AI agents and powerful quantum computers. 

While AI might not replace humans entirely, those who embrace this powerful tool will be the future leaders in research. The possibilities are endless when highly developed AI merges with quantum computing, wielding the knowledge of entire scientific fields. This vision is closer to reality than we think. The key is to consider AI as a friend rather than a foe and learn to leverage its strengths alongside our unique human capabilities. 

The Journey Into Industry

Dr. Ashok Kumar, a resident of Mumbai, India, is a distinguished scientist with a passion for pushing boundaries in drug discovery. He holds a PhD and the prestigious title of Fellow of the Royal Society of Chemistry (FRSC).

Following his postdoctoral research with Professor Sir John Cornforth, a Nobel Laureate at the University of Sussex, Dr. Kumar has co-authored over 35 peer-reviewed publications in international journals and co-invented over 100 patents. Since September 2005, he has led the charge as President of the research and development centre at IPCA Laboratories Limited.

Dr. Kumar's philosophy, inspired by Richard Feynman, is to "Study what interests the most in the most undisciplined, irrelevant & original manner possible and think what others have not thought." This approach fuels his expertise in new drug discovery, biotech/biosimilar R&D, intellectual property creation, and innovative process research within the field of chemistry.

A captivating motivational speaker, Dr. Kumar has received numerous awards and delivered countless talks, sharing his knowledge and igniting curiosity within the scientific community.