Founders: Trisha Chatterjee & Aridni Shah
Healthcare today is moving towards targeted therapies as they are more effective and have lesser side effects compared to the traditional treatments. Antibodies are an emerging candidate for targeted therapy and have shown promise for a multitude of diseases ranging from chronic illnesses like cancer and autoimmune diseases as well as viruses like Ebola, Respiratory Syncytial Virus (RSV). For instance, several monoclonal antibody therapies (mAbs) have been granted Emergency Use Authorization by FDA for treatment of Covid-19. There has been an exponential rise in the demand for these therapies in the US and research has shown monoclonal antibodies to be effective against Covid-19 and can potentially reduce the risk of hospitalization or death by at least 70%.
Antibodies is a promising class of therapeutic drugs that are essentially lab produced proteins that mimic the human immune system’s response to pathogens like viruses, cancer cells, bacteria, etc. They bind to their targets/antigens with high specificity, reducing side effects and improving outcomes. The global monoclonal antibody therapy market size is projected to grow from $158B in 2020 to $452B in 2028. Over the past five years, antibodies have become the best selling drugs. For example, Abbvie’s Humira (derived from “human monoclonal antibody in rheumatoid arthritis”) generated $20B in revenue in 2020. While antibody discovery and engineering has evolved significantly since the first FDA approved antibody drug in 1986, there is still a lot of growth potential. There have been only ~100 FDA approved therapeutic mAbs in more than three decades.
So why is that the case? Unlike small drug molecules (which form 90% of the drugs in the market), antibodies are large molecules/biologics, which are much more complex. The current processes for discovery and screening of antibodies are primarily lab driven. Antibody discovery companies typically have a large base of antibody candidates (100-1000s) which have to be then screened through lab experiments to determine whether an antibody binds well to an antigen, referred to as “binding affinity”. The best set of candidates identified through this process are then selected for clinical trials. These processes can take more than a year and cost millions of dollars. The clinical trial approval percentage can be as low as 10%.
Figure: Traditional Antibody Discovery & Screening process
Images from Front. Microbiol
Automating this process through computational methods can fast track the discovery. The goal is to be able to design and identify antibodies that bind the best with a given target antigen. High accuracy predictions can be obtained if the antibody-antigen interactions in 3-D space is taken into consideration. However, experimental methods of 3-D structure prediction are time-consuming and difficult. On the other hand, predicting the 3-D structure of antibodies is challenging since these molecules are complex and more difficult to characterise than traditional small molecule drugs, given their variable loop structure. Also interface prediction is difficult because antibodies are ad-hoc binders.
This is where immunitoAI’s value proposition lies. Using its biology-first hypothesis and AI based implementation, the company’s platform examines the spatial conformation of the complex and uses proprietary feature engineering and Deep Learning models to identify the most suitable antibodies that have high specificity and sensitivity. The platform predicts 3-D structures of the complexes followed by extraction of chemical bonds and solvent interactions, amongst others, resulting in high accuracy binding affinity prediction. For a given target antigen, the platform ranks and delivers a smaller set (10s) of antibody candidates for further lab screening.
Figure: immunitoAI’s approach to Antibody Discovery and Screening
imRANK, the first product under development by the company, screens and ranks the best antibody candidate while imEVOLVE, the next product in the pipeline, will suggest mutations in suboptimal or failed antibodies to develop improved candidates ready for clinical trials. immunitoAI’s platform has the potential to make the antibody discovery process 2x more efficient, 60% faster and 50% cheaper. While the focus is on antibody screening and improvement in the short term, the company plans to develop their own antibody candidates in the mid and long term.
Aridni and Trisha founded immunitoAI in November 2020. Aridni brings a strong domain expertise through research background in molecular biology while Trisha brings years of applied AI experience and together they bring a unique combination of biotech and AI. We first met them a few months back and were impressed with their skills and passion for solving a high impact problem.
While this is our first investment from our recently launched Fund II, this is also our first investment in core biotech. While the risky, regulated, capital intensive and long gestation nature of biotech can make it unattractive to a lot of investors, we believe it is a high risk and high return sector that is a true reflection of the power law of returns.
We are proud to see such deeptech innovation being attempted out of India and can’t wait to see it touch lives around the world.