.png){kind=logo}
- Insights
- Cryptocurrencies
- Stocks
- White Papers
- Industry
- Geography
- Insights
- Cryptocurrencies
- Stocks
- White Papers
- Industry
- Geography
Testing blood cells are not done in a proper way in countries like India and hence a lot of people get affected by unwanted diseases. Roughly 25,000 newborns in India every year are at risk of preventable mortality or lifelong morbidity due to metabolic disorders when there exists a test to detect their condition and treat it. So, it has become absolutely important to come up with solutions that would minimize such use cases of market failure. Algorithmic Biologics has come up with a product called Tapestry, which is a molecular search technology delivered on the cloud. Molecular tests are great at finding a needle in a haystack whereas Tapestry is great in situations where there are many haystacks, and few of them have needles. Analytics Insight has engaged in an exclusive interview with Manoj Gopalkrishnan, Founder of Algorithmic Biologics.
If you receive a blood transfusion in India, you are 3000 times more likely to pick up HIV from the blood transfusion than if you receive it in the USA. This is because India does not use the more expensive Nucleic Acid test to test donated blood units. Roughly 25,000 newborns in India every year are at risk of preventable mortality or lifelong morbidity due to metabolic disorders when there exists a test to detect their condition and treat it. The food we consume is routinely found contaminated with pesticide residues or microbial growth or antibiotics to unacceptable levels, causing health problems, when tests exist to detect such contamination.
These are all market failures: the test exists but is not being used. The simple reason for this market failure is cost — the tests are unaffordable. Our mission is to solve these market failures by making gold-standard molecular testing much more affordable.
Our journey began with Covid19. We realised how countries in east Asia were using Covid19 testing as an effective way to flatten the wave. In India in contrast, we were rationing testing because tests were a scarce resource. We asked ourselves, how would a computer scientist solve this problem? After all, if 1000 people are being tested, and say only 10 people are positive, we only need to find the 10 positive individuals. Can there be a better way that does not require 1000 tests?
We used techniques from AI to come up with our product Tapestry, which is a molecular search technology delivered on the cloud. Molecular tests are great at finding a needle in a haystack whereas Tapestry is great in situations where there are many haystacks, and few of them have needles. We are able to identify the haystacks with needles in very few tests in a single round of testing. Not just that, we can tell how many needles are in each haystack without any additional testing.
Many people have been trying to solve this problem for many years, but we have cracked it, building not just great algorithms based on fundamental scientific modeling, but also the last-mile solutioning required to make this practice without any additional capital expenditure. This has led to recognition in the international scientific community with prominent scientific news publications like Nature, American Mathematical Society, Institute for Electrical and Electronics Engineers, and Society for Industrial and Applied Mathematics all spotlighting our innovation. We were finalists in the XPRIZE for Rapid Covid Testing, the only team from India to reach the finals, and one of only ten teams in the finals from a field of roughly 300 teams worldwide.
We took the opportunity the Covid19 crisis offered to take the technology commercially with a regulatory nod from DCGI and obtained a CE mark from the European Union as well. We have tied up with multiple diagnostic labs including Thyrocare, one of India's leading diagnostic lab chains. We have tested more than 25,000 samples and demonstrated savings in the field. We are in pilots with market leaders in India for blood bank screening, newborn screening, food safety, seed research, and more use cases.
Our team! You see, it is hard to build AI for molecular testing. You need to know the science. Machine learning is no cakewalk either because this is time-series data and often there isn't a lot of data. You can't just import a library function from PyTorch or TensorFlow, you have to really know your stuff on both sides to innovate in this space. Even a multidisciplinary team will not cut it, you need individuals who understand both sides so that the meaning can crystallize in their brains, and new ideas can emerge. Such individuals are hard to find.
My own journey has been a transition from being a scientist having spent 18 years in top research universities in USA and India to beginning this entrepreneurial journey. My area of research is molecular computing, which combines molecular science and computer science. I am equally comfortable writing code, running experiments in a molecular biology lab, or doing math. This turned out to be a big advantage when I sat down to tackle these problems and gave us our edge in the XPRIZE. So when I started hiring, I went looking for other individuals with those traits, and we have put together a team of a dozen folks of whom 5 are PhDs. Each individual is interdisciplinary and can make sense of both sides of the equation, molecules, and algorithms.
We have folks on our team who are based in London and Los Angeles and are working full-time with us because they are so excited about the problems and the impact possible. And all this hiring has happened in a tough hiring climate, when the great resignation was going on, on an early-stage startup's shoestring budget. It's sheer madness, I can't explain it any other way!
I think Covid has been a game-changer culturally. No one thought of molecular testing, now it is something commonplace. With that mindspace comes a cultural shift. Next time you are unwell, won't you want to know exactly what is wrong with me? Won't you want more information, and want to take ownership of your health and wellness in new ways, now that you realize the power of the invisible microscopic scale over your life?
So more than how it has evolved, which has been substantial without a doubt I am looking at what is coming next. The millennial and Gen Z want more data, wants more control. And that data, that sense of control is going to come from being aware of what is in your body and what is going into your body and what is in your environment.
We are a platform technology. There are many situations that can be expressed as molecular search. Many apps can be written on top of this platform, of which Covid19 was only one example. Our partners are the real heroes, they are trying to innovate to create drought-resistant seeds, provide safe blood for blood transfusions, make newborn screening universally accessible, and many other use cases. They have deep expertise in these areas and are committed to bringing this change about. Often it is a matter of finding some way to offer the solution at a lower price to make that market. That is where our technology comes in.
For example, one of our partners is a leading blood bank in the country. We are in pilot with them. Another is Mahyco, India's largest seed company. Thyrocare, one of India's leading diagnostic lab chains, has been a big supporter of our Covid19 screening efforts, and we are identifying other areas where we can work together.
A customer wants a molecular technology that just works. The biggest pain point is not knowing whether the test is reliable. Getting the test right requires expertise in molecular science, which is a rare commodity. There are so few experts! Tapestry has a side benefit apart from search: it allows every lab to identify problems early. In telecommunications technology, this has been known as "error detection." Such a simple technology, but it has not been available at scale in molecular testing. The next frontier from here would be if a customer can verify that the molecular test is correct, without needing to trust the test.
Let us consider the food and beverages industry as an example. Today the reality is everyone trusts somebody else to act ethically. The customer trusts the packager, the packager trusts his many suppliers. There is almost nothing in the way of quality control at the supply end. More often than not, many small suppliers are feeding into the supply chain. Even if one supplier slips, either knowingly or unknowingly, the end product becomes unacceptable. But there is no way to trace back and find out which supplier was at fault.
With advances in AI and molecular technology, this is a point of shift for the entire industry enabling things that were never possible before, like testing at the supply-side and source tracing. The organizations that are able to access these advantages will be able to assure customers of higher quality.
Or consider dairy. There are so many new-age D2C brands selling A2 milk and milk from specific Indian cow breeds. They charge a premium, but there is very little that customers can verify. Ensuring quality is only half the problem, assuring quality also needs to be solved so that the customer perceives the higher quality that you have created.
The first challenge to confront for any AI company starting off is that building models need data, and where are you going to get the data? So you need to have a data bootstrapping strategy, where you need to be smart about how you are going to get your product off the ground in a situation when you don't have much data. We did this successfully for Covid, and we are following the same strategy going ahead. Many out-of-the-box solutions like deep learning rely on big data, so they are often not good technologies to start with. Once the product is commercial and scaling up, data accumulates very quickly and these technologies come into play.
At Algorithmic Biologics our approach is patented and proprietory. So, even though everybody seems to be talking about engineering biology, you can't really do engineering unless you can build a simulator. Without that, it is pure trial and error, and the risk of failure will limit the scale of what you can achieve. Our team is really keen on doing this hard work of building simulators, we are investing in this direction.
