Note: I’m neither a chemist nor a biologist (just an ML researcher who used to work on problems in CompBio), so I welcome feedback from people with more knowledge and experience in these fields. This proposal is too short and underdeveloped to qualify for the Cause Exploration Prize, but I would be delighted if someone else takes this and turns it into a real submission to the contest.
In the development of therapeutics, it is typical to screen many compounds (or targets) through in vitro assays, and then to assay the successful candidates through more expensive in vivo experiments. False positives are extremely common; most candidates which pass in vitro testing are expected to fail in vivo. Furthermore, it is sometimes observed that candidates which fail (or are relatively less promising) in vitro are successful in vivo. Such discrepancies are not surprising, since cell culture conditions and even the cell lines themselves (which often have been passaged repeatedly) are vastly different from living tissue. Because the differences are so vast, and the discrepant outcomes are so expected, such discrepancies are rarely deeply investigated.
However, aggressively minimizing such discrepancies ought to be an area of focused research to accelerate therapeutic development. The reliance on in vivo experiments, in addition to raising ethical quandaries, means biology is a labor-intensive field with much monotonous, unpleasant work. The inhibits many bright, sensitive minds from entering the field. It also means that lab automation efforts, which really only accelerate in vitro assays, have limited upside due to Amdahl’s Law.
Furthermore, understanding these phenomena at a precise, mechanistic level may unlock new insights into biology. The development of cell lines and cell culture media in the 20th century helped improve our understanding of the biological importance of various enzymes and minerals. However, the era spanning from the time of John F. Enders to Richard G. Ham has passed. Young, ambitious, and curious scientists no longer dive into this now-sleepy field. Current lines and media are generally considered “good enough,” with the exception of popular current research efforts into 3d media. However, it is very possible that a broad revisitation of this field, taking advantage of developments in gene editing and computing power, would yield novel insights into biology.
One possible avenue of research would be to focus on specific known cases where in vitro and in vivo experimental results differ, and then dive deeply into understanding the gap. Cases where compounds succeed in vitro but fail in vivo are abundant, of course. So it would be important to pick particular cases which are interesting in themselves, and which could reveal generalizable insights. But more interesting and impactful would be to identify compounds which succeed in vivo yet fail in vitro. Such cases are harder to find, but understanding them would give us visibility into the (compound, tissue, disease) triplets that are successful yet current in vitro screening methods are unintentionally screening out.
Cause: research to narrow the gap between in vitro and in vivo experiments
Note: I’m neither a chemist nor a biologist (just an ML researcher who used to work on problems in CompBio), so I welcome feedback from people with more knowledge and experience in these fields. This proposal is too short and underdeveloped to qualify for the Cause Exploration Prize, but I would be delighted if someone else takes this and turns it into a real submission to the contest.
In the development of therapeutics, it is typical to screen many compounds (or targets) through in vitro assays, and then to assay the successful candidates through more expensive in vivo experiments. False positives are extremely common; most candidates which pass in vitro testing are expected to fail in vivo. Furthermore, it is sometimes observed that candidates which fail (or are relatively less promising) in vitro are successful in vivo. Such discrepancies are not surprising, since cell culture conditions and even the cell lines themselves (which often have been passaged repeatedly) are vastly different from living tissue. Because the differences are so vast, and the discrepant outcomes are so expected, such discrepancies are rarely deeply investigated.
However, aggressively minimizing such discrepancies ought to be an area of focused research to accelerate therapeutic development. The reliance on in vivo experiments, in addition to raising ethical quandaries, means biology is a labor-intensive field with much monotonous, unpleasant work. The inhibits many bright, sensitive minds from entering the field. It also means that lab automation efforts, which really only accelerate in vitro assays, have limited upside due to Amdahl’s Law.
Furthermore, understanding these phenomena at a precise, mechanistic level may unlock new insights into biology. The development of cell lines and cell culture media in the 20th century helped improve our understanding of the biological importance of various enzymes and minerals. However, the era spanning from the time of John F. Enders to Richard G. Ham has passed. Young, ambitious, and curious scientists no longer dive into this now-sleepy field. Current lines and media are generally considered “good enough,” with the exception of popular current research efforts into 3d media. However, it is very possible that a broad revisitation of this field, taking advantage of developments in gene editing and computing power, would yield novel insights into biology.
One possible avenue of research would be to focus on specific known cases where in vitro and in vivo experimental results differ, and then dive deeply into understanding the gap. Cases where compounds succeed in vitro but fail in vivo are abundant, of course. So it would be important to pick particular cases which are interesting in themselves, and which could reveal generalizable insights. But more interesting and impactful would be to identify compounds which succeed in vivo yet fail in vitro. Such cases are harder to find, but understanding them would give us visibility into the (compound, tissue, disease) triplets that are successful yet current in vitro screening methods are unintentionally screening out.