Daniel Fürth

(New thesis) AI-ification of R&D proteins There’s been an emergence of new papers on computational antibody, peptide, nanobody, and and enzyme design. While these are all relevant for new therapeutics, there’s massive market opportunity in R&D reagents. (This thesis was started by a conversation with @an1lam.) Research and development in companies and academia stems from our exploitation and synthesis of new proteins (namely, antibodies and enzymes). From staining tumor samples, to copying DNA with polymerases, to cutting DNA for cloning, proteins are workhorses of biotech. Not only this, but proteins are the commodity most likely to spoil because of their propensity to denature at room temperatures and through freeze/thaw cycles. Not only this but the and cost to make and purify means proteins are likely the most expensive reagent in experiments. If you’ve ever worked in a research lab you know that you always need to keep the antibodies and enzymes chilled because they cost hundreds to thousands of dollars and are the first failure point of experiments. Given the proliferation of antibody and enzymes design papers, it’s logical to apply the gains in those methods to R&D proteins. Namely, thermostability, miniaturization of antibodies for cheaper manufacturing, higher fidelity signals or specificity, all of which can be improved with the right assay <> AI loop. Not only this but @plasmidsaurus teaches us that good user experience and better products can lead to massive customer pull and rapid uptake. This matters commercially because these companies garner therapeutics-value acquisitions and market caps. While these have historically taken decades to reach prominence, the decreased cost and speed of development of new proteins could change the incumbent dynamics. - $5.25 billion acquisition of antibody and reagent maker, Biolegend by PerkinElmer - $5.7 billion acquisition of research antibody company Abcam by Danaher Bio-Rad Lab market cap is $7.3 billion - Smaller acquisition of companies such as Novus Biologicals by Bio-techne for - $60 million show a skew of outcomes possible Other thoughts: - Team will have to thread the needle of highest margin and lowest technical difficulty - Team likely won’t have to build their own models initially but the wet lab <> ML feedback loop will be crucial in rapid optimizations - Companies here could expand margins by layering biomanufacturing unlocks to produce and make R&D proteins at lower cost, in less time Check out the thesis below on the @CompoundVC Thesis Database -

Come work with us! We're hiring a postdoc to lead a new project on human social learning in threatening and volatile contexts. More info & apply: ki.varbi.com/en/what:job/jo…

nature.com/articles/s4146…