Dr. Jo Varshney was born to a pharmacological pedigree in India. Being introduced to the world of drug development and research at a young age and an ethic of competitive achievement, she developed a passion and understanding for science and technology much earlier than most.
Today, she is the Founder and CEO of VeriSIM Life, and the inventor of VeriSIM Life’s BIOiSIM core technology.
A virtual drug development engine that harnesses the power of traditional statistical modeling along with artificial intelligence and machine learning, BIOiSIM enables pharma and biotech professionals to bring more lifesaving and cost-efficient therapies to patients faster than ever before.
Over the past several years, Dr. Varshney has front-led numerous collaborative engagements with a host of pharmaceutical clients, government agencies, academic/medical institutes and industry innovators.
She is a celebrated leader and has received numerous accolades, as well as delivered keynotes at several conferences.
Prior to starting VeriSIM Life, Varshney gained her doctorate in Veterinary Medicine (DVM) and holds a Ph.D. in Comparative Oncology/Genomics from the University of Minnesota, as well as graduate degrees in Comparative Pathology from Penn State, and Computational Sciences from UC San Francisco.
In a compelling conversation with Breaking Cancer News, Dr. Varshney, shares her latest cancer breakthroughs, including her work in Artificial Intelligence (AI) in cancer and other new technologies that are changing the cancer landscape as we know it.
Q – How and why did you choose this road with all its challenges?
I was raised in a competitive household in India, where I was exposed to the pharmaceutical business pretty early on. Seeing at a very young age the various steps and inefficiencies in the development of a drug really got me obsessed about improving the process.
I took programming courses at the age of 8. I also wanted to understand the purpose of animal testing in drug development to discern how that impacts a drug’s clinical success. This is what led me to become a veterinarian.
My obsession persisted, and I wanted to explore how computer programs could be leveraged to understand differences between humans and animals. In 2010, I came to the United States to study comparative pathobiology/oncology, genomics and bioinformatics.
Over the next six years and during my PhD, I worked on several diseases including cancer that enabled me to develop novel biomarkers and 3D organ models, conducting NGS- small RNA sequencing and whole transcriptome analysis.
My research revealed a troubling reality about helping patients. I realized how difficult it is to translate from the lab to clinical trials and thought there must be a better way. So, I studied computer science to use machine learning, mathematical models, and data to find a better, faster way to bridge the gap between lab and patients.
My first version of creating a virtual animal-to-human drug simulation was comparable with real patient data, and I was quite astonished! This was a pivotal moment for me. It gave me confidence that with the right technology and scale it would significantly impact the patients positively.
I entered a Google hackathon to build out the concept and won. These moments helped me take a bet on myself, and I started VeriSIM Life shortly thereafter.
Looking back, 10 years later, I feel fortunate to have undergone many challenges to pursue a nontraditional career. Also, I want to emphasize the value of becoming obsessed with a problem. Several times before the field of artificial intelligence was recognized as a credible technology, I had to shut out the naysayers and really believe in what I understand and how we can solve the drug development challenges.
I wouldn’t have created this first in-kind business without having learned the complexity of drug development firsthand.
Q – When and why did you decide to focus on AI?
As you know, more than 95% of drugs fail in late-stage clinical trials. You may not know that the biggest reason is that many drugs tested in animals don’t show the same results in humans. This poor translatability between animals and humans is the main drug program killer and reason for extensive time and drug costs!
When I was studying in the early 2000s, artificial intelligence (AI) was only starting to be pursued seriously in commercial applications. However, I could see the potential for its use, especially in learning from existing knowledge and predicting patterns.
As you know, biology is extremely complex. I thought that if I could codify critical aspects of human and animal biology, the animal-to-human model translation problem would be resolved. The great thing is, there are already many knowledge models that could really help in learning biology.
So, I took several machine learning courses and started playing around with this idea. Pretty quickly, I found that fusing AI with these models returned amazing results. This was the start of the formation of VeriSIM Life!
And now we’re able to do things well beyond my small-scope approach. We now utilize mathematics, physics, biology, and chemistry to help AI learn different relations and ensure that it doesn’t go crazy. It knows a lot about how the body works, how chemistry works, what rules to follow, and where to look for novel patterns.
The system connects the dots between physiology, chemistry, genomics, proteomics, metabolomics, species variability, and more. So now we can answer more questions about the clinical translatability of small molecules, biologics, peptides, and formulations.
Here’s a great example of how it helps in practical scenarios: Earlier this year, we finished a client project that centered on a molecule for the treatment of metastatic melanoma. The client’s animal experiments suggested that instead of the actual drug, one of its metabolites was creating a positive drug response.
We were able to predict which one amongst the thousands (3,900, to be precise) of metabolites. This would not have been possible without our approach and it would have risked either pausing development or pursuing clinical trials that would have failed. And now they’re moving ahead with the right molecule.
That is the power of our approach. I am really glad I took the chance and created this company, and every day I am honored to work with brilliant minds who have made my dream come true and who truly care to see patients getting treatment faster.
Q – What are you working on now, and what are the challenges?
I’m especially interested in where we will end up with generative AI. Generative AI technology is what you may have seen at work in tools like ChatGPT. It has the ability to take a “seed”, like some text, and create from it a very complex structure, like more text or an image or video. In our case, the complex structure we’re creating is a completely new drug!
We recently unveiled a new platform component called AltasGEN that is helping design drugs with translatability built-in from day one. AtlasGEN is already being utilized to design drugs for very difficult targets.
I’m very optimistic based on how quickly we’ve been able to generate and select promising candidates. To me, time to market is the challenge that we want to solve with our approach.
I am hopeful that with this near-perfect drug design and testing loop we can speed up experimental validation and compress the entire R&D cycle even more.
Q – Please explain how your platform actually works and what impact it could have on the cancer landscape:
Basically, our platform combines knowledge from the early stage of research to late stage of development, which extends to clinical trials, and generates a score we call the Translational Index.
It’s inspired by the concept of a credit score. For example, your credit score indicates your financial health, and based on that you make the right financial decisions.
The Translational Index score takes into account different aspects of drug development based on the disease, the drug’s chemistry, target, etc. So when you get a score, it tells you how successful it will be in clinical trials and what you can do to improve the odds.
Such an approach increases AI explainability and ultimately speeds the drug development and approval process, as well as reduces reliance on animal testing.
We’ve used our platform to accelerate our own investigational new drug asset for rare diseases. In less than two years, we’re putting together an IND package for our drug candidate.
Our entire drug development journey has cost us less than a million dollars of investment which is 50% lower than a traditional approach.
Q – If successful, what could this mean for the future of cancer treatment?
If we can help drug companies get more good therapies into patients’ hands by cutting development time and costs and improving clinical success, that’s a huge win.
We also think we can help create more personalized treatments such as making children’s oncology treatments safer and more tolerable, an area of critical need for young cancer patients.
I’m proud to be a part of a new solution that addresses so many unmet patient needs.
