The COVID-19 pandemic triggered many ideas among the tech professionals, and molecular programming for the regular RTPCR test is one of them. Algorithmic Biologics, a deep-tech startup, is building algorithms for molecular programming. The company was founded by Dr Manoj Gopalkrishnan, a faculty member at IIT Bombay.
Dr Manoj holds a B Tech from IIT, Kharagpur, a Ph D in Computer Science from the University of Southern California Viterbi School of Engineering and has taught at the University of Southern California, Duke University, and IIT Bombay before founding Algorithmic Biologics.
His innovation, Tapestry, is a single-round quantitative method for large-scale molecular testing that provides substantial time and cost savings over traditional RTPCR tests. It is validated on over 15000 samples at reputed Indian and international labs. It is the only indigenous Indian technology to reach the finals of the XPRIZE for Rapid Covid testing competition.
Analytics India Magazine interacted with Dr Manoj to get insights into molecular programming and his research.
AIM – Share your journey of launching Algorithmic Biologics (AB); how did the idea come about?
Dr Manoj – I am a professor-turned-entrepreneur. I have been a researcher for the last 18 years in molecular computing and held faculty positions at TIFR Bombay and IIT Bombay for 14 years, with expertise in both molecular science and computer science. My research has essentially been focused on how to build smart soup: chemical systems that can perform computation.
When Covid-19 came, I realised that some of these ideas could transform molecular testing. Bringing these ideas of AI and Molecular Computing to Molecular Testing could enable better health for all, not just in the domain of Covid-19 screening where we began, but in various other areas, including newborn screening, food safety, and discovery assays in life sciences, biotechnology, and pharmaceuticals. This was the vision behind the company.
Covid-19 was the specific trigger. It allowed the right collaborations to fall into place. As a result, we went through a period of accelerated innovation that is unprecedented in my experience. Everything was fast-tracked, whether the innovation or the software development, or even the trials permissions, regulatory approvals, and commercial adoptions. We managed to get DCGI nod and CE approval and go commercial with lab partners like Dhiti Omics and leading diagnostic lab chains like Thyrocare.
AIM – What challenges did you face while launching AB, and how did you overcome them?
Dr Manoj – In our product innovation, we made the mistake of associating our idea with “pooling” early on. Unfortunately, pooling is an idea that people think they understand, so they stop listening at that point. They imagine that all the negatives of pooling will carry over to our method.
In hindsight, we are not a pooling technology at all. We are a molecular search technology. Think about Google search on the internet. Searching for words in documents was already an existing technology before Google came about. What Google solved was that when there were many different documents — for example, the entire internet — it told you which websites contained your words of interest and ranked these websites based on relevance. Analogously, when there is one sample, testing for a target in that sample, like the Covid-19 virus, is what molecular testing is good at. Our technology is great when there are many, many samples. We can find the few samples that contain the target in a single round with a very small number of tests in a single round of testing. In addition, our results are multiplexed and quantitative.
To make this concrete, we regularly give results for 1000 samples with only 100 tests and catch ten or so positives in a single round of testing. And we are doing this commercially with our lab partners with great turn-around times and concordance with individual testing. Our algorithm receives the lab testing results and does the solution on the cloud. There is no magic here; you can understand every step of what we are doing.
There are three key innovations. The first is the pattern of pooling. For this example, each test is performed on a pool of many samples, around 30 for this case. Each sample participates in three different pools and is tested three times. We want this pattern to have good mathematical properties, which will help in the algorithmic step we perform later. At the same time, we want this pattern to be easy to perform accurately and fast in the lab. We have solved this problem and filed IP for it. The second innovation is that, unlike other methods that have used only whether the test is positive or negative, our algorithm can use the quantitative information returned by the test. This gives our method the power to handle a larger number of positives. The third innovation is a noise model based on science-based modelling of the underlying chemistry fed into our AI. This makes our method robust.
Several teams have been trying to solve this problem for the last two decades and have fallen short of a practical solution. We have succeeded perhaps because of our transdisciplinary expertise, which enabled us to combine innovative algorithms with the last-mile solution required to make this technology practical without any additional capital expenditure. Prominent scientific news publications like Nature, American Mathematical Society, Institute for Electrical and Electronics Engineers, and Society for Industrial and Applied Mathematics spotlighted our innovation. We are finalists in the XPRIZE for Rapid Covid Testing. This recognition from the international community helped us get people’s ears. We were able to demonstrate the product to them in a way that convinced them and led to adoption. Today our lab partners have become our biggest champions and supporters. We are working together to explore other use cases, and they are promoting us on their own with other labs in their network.
AIM – Who are your competitors, and how is AB different from the products and services of your competitors?
Dr Manoj – The molecular testing industry is about $60Bn, of which a $10Bn segment is “scale multiplier technologies.” This segment is seeing rapid growth at 5X the industry average. We are playing in this segment. The offerings that bring scale to molecular testing are automation and integrated fluidics, novel chemistry and biologics, and algorithm-based efficiencies supported by the software. We are in the third category, which has the unique advantages of being low capital cost, platform technology independent of assay and machine, and easy solution delivery through the cloud.
This space is greenfield; less than 40 startups are using any kind of AI for molecular testing. We believe this segment is ripe for a huge growth spurt like we have seen in AI-based medical imaging in the last five years. Our patents are going to be a big advantage in the future. There is also a lock-in possible if we can capture the OEM market and integrate with OEM solutions.
AIM – Please explain in detail how you are analysing biological data (DNA, RNA, protein, etc.)?
Dr Manoj – We are not analysing biological data; we are helping collect such data. This is an important distinction. We are not downstream of existing molecular tests; we wrap around them to make them more powerful.
That said, there is a data play here that will emerge. Over a period of time, we will have access to substantial biological data, and new opportunities will arise from that. But at this stage, that is not our focus. We are focused on making molecular testing smarter.
AIM – What is the tech stack involved in the analysis?
Dr Manoj – We do modelling of the underlying molecular assays based on fundamental science and convert these models to software. We use these models to drive our inference engines and decision-making. The whole solution is hosted on the cloud and delivered via our web app.
AIM – Tell us about Tapestry in detail and how it is different from RTPCR testing machines.
Dr Manoj – Tapestry is a software solution. Our software helps labs pool many samples into a small number of pools. They test these pools using their standard lab protocol and their standard assays and machines. The results of this testing are uploaded to our web app. After the cloud-hosted solve, the solved results are displayed within minutes on the web app.
AIM – How is AI used in the field of biotech? How is India progressing in this field?
Dr Manoj – Biotechnology is understood as the domestication of the cell, but I want to talk about something bigger. If you draw an analogy to the domestication of the horse, then molecular computing can be compared to the automobile. It is about technology inspired by biology, similar to how the aeroplane was inspired by bird flight.
Biotechnology, in this wider sense, is going to be the story of the next three decades. Many VCs are bullish on this opportunity; IndieBio has estimated this as a $100 Trillion opportunity, a stunning number.
A new kind of programming is emerging, the programming of molecules. The styles required for this kind of programming need to consider the ever-present stochasticity at the molecular scale. This is where AI has a big role to play since AI is a style of programming that is very robust to noise. This AI running on reaction networks will complement AI that runs on silicon computers. Having the ability to program both inside and outside a test tube will be a key strategic advantage going forward.
India has a wealth of scientific talent, technology talent, and pharma-biotech industry expertise. We are a very fertile ground for this kind of innovation. The biggest challenge is to promote individuals who are transdisciplinary and can seed this kind of innovation. Multidisciplinary teams are great at taking innovation to product, but history shows that integration of ideas and meaning-making are required for this kind of innovation. This happens much easier if one person knows both sides.
AIM – What are your future plans for AB, and how do you plan to pursue it?
Dr Manoj – We are focused on applying Tapestry to various use cases that will help improve health outcomes at scale. We are making accurate molecular testing available at scale. We will continue engaging with our lab partners and domain experts as we seek to grow through channel partnerships. We are also interested in integrating Tapestry with OEMs for specific use cases.