As the year draws to a close, we look at how AI and software have transformed what wasn’t exactly ‘a happening space’ just a decade back. With every new year, big strides are being made in the Biopharma world. Nearly eight years ago, when we founded Elucidata, not too many investors or customers were convinced of the impending paradigm shift. This was probably because Biopharma has traditionally been hard to disrupt from the outside. Few companies have been able to penetrate this space.

Fast forward and it is a completely new ballgame. Early-stage companies are raising large rounds. Tools catering to various stages of the scientific workflow are becoming common (see chart below). Applications of AI are already mainstream; a clash of the titans, Alphabet and Meta, is raging. Meta AI’s new protein structure prediction model ESM2 is, in some ways, an upgrade from Alphabet’s AlphaFold. Disruptive innovations like these have invigorated protein prediction models, both in their development and use.

Now, initiatives such as AlphaFold are going open-source, a dream come true for chemists! This is just one example of how monumental AI is in advancing life science research. Data mining is another minefield where AI is reshaping the research landscape. Models can swiftly sift through literature to parse relevant information about experiment design or even to identify novel targets.

There is a whole spectrum of opportunities which can come with the adoption of software tools in R&D. And there have been significant efforts by many new companies in the Electronic Lab Notebooks (ELN), Laboratory Information Management System (LIMS), and Lab Automation space.

Experimental data is still stuck in excel sheets and is almost never organized at one place. A few startups, like Benchling, that solve this problem are being widely adopted as defaults. The scenario was not the same just a couple of years ago and companies would build in-house tools to suit their workflows. However, in the past decade, this has changed rapidly as shown by this chart of software tools in discovery. Even imagining such a stack was unfathomable.

So what does all this mean for the upcoming year and the world at large:

• Increased use of software and tools for automation. There is a tangible shift from problem solving in-house to adopting collaborative approaches for better results.
• Expedited evolution of cloud infrastructure owing to the large and rapid data production.
• Leveraging AI as a differentiator, especially by smaller companies, since it brings a massive competitive advantage. The use of AI has largely been limited to niche full-stack platform companies such as Atomwise. However, it has become more available, usable, and adoptable across the biotech industry more recently.
• A push to enable the availability of data - new directives such as the OSTP guidelines are urging researchers to make their data public. Read our recent post here about the new NIH policies mandating data release.
• Data harmonization and standardization may become a major problem owing to generation of large volumes of data and its availability. On the bright side, Polly can help you with this to accelerate early stage drug discovery by 75%.
• Conclusive data will hopefully start to emerge on how AI is aiding drug discovery and making it faster/cheaper. The conclusions we have drawn are based on limited evidence, more data is required to substantiate the success we’ve achieved so far, i.e. multiple molecules found using AI (at some stage of the process) are now in the trial stages.

In summary, both in terms of AI applications and software adoption, there has been a significant shift in 2022. Even though the market is tough, more of the same will continue. Exciting times are in store for sure, looking forward to more advancements in 2023 and beyond!

‍

This post was originally published in Polly Bits- our biweekly newsletter on LinkedIn.