“Is Target-Based Drug Discovery Efficient? Discovery and “Off-Target” Mechanisms of All Drugs?”
important recent article by
@arash4sadri on what I agree is the main flaw of how the vast majority of drug discovery is done today
pubs.acs.org/doi/10.1021/acs… (full text preprint -
osf.io/preprints/osf/bxwcr)
tldr: the "target-based" approach - based on "hitting" a single protein (gene) - remains the dominant drug discovery approach for decades despite discovering less than 10% of all drugs
on top of this, only a fraction of the 10% target a single protein. the mechanism of action (MoA) of those 10% is often polypharmacy - hitting multiple targets and often what are discarded as "off-targets"
thus, so much effort is spent on the target-based approach and yet it produces so few drugs
the waste is so large it’s been documented by
@JackScannell13 and is derisively known as “Eroom’s law” - which states it takes longer and costs more to develop drugs each year. current estimates are nearing a decade of time and costing > $1B per new drug
back to the article: target-based drug discovery contrasts with the much more successful approach of phenotypic screening, which has discovered 90% of all drugs. famous examples include aspirin, penicillin, and for the longevity folks - metformin, bisphosphonates (see further below), and others
some background on target-based vs. phenotypic screening: one can think of target-based as gene-centric. another way it's been described is "rational" drug design. this implies it is irrational to design a drug any other way 🤔. it has become the dominant mode because often people start with a disease and try to find a cause. The general approach is as follows: disease->target->drug. this makes sense but is overly reductionist as we and Arash argue. another way of saying this is that researchers have strong biases at each step in going from disease->target->drug
in contrast to target-based, phenotypic screening can be thought of as drug-centric. Arash makes several arguments based on information theory on why target-based is inferior to phenotypic screening, but the bottom line is phenotypic screening’s cause-and-effect is much more tractable
still the problem with phenotypic screening is one doesn't know how the drug has the good effects it does. thus, there is no path to improve the starting drug. and one needs this both to make a better drug but also to build intellectual property around it such that one has enough incentive to pursue the hard and expensive work of getting the drug to the market
@bioio_tech we think we’ve developed the best of both worlds with our MOAT (mechanism of action technologies) platform. our approach combines phenotypic screening and target identification in a single experiment
how? we start with a drug that has the desired phenotype of interest (e.g., it's already a proven anti-diabetic) and we do genome-wide screening using gene editing technologies such as CRISPR to identify which of the 20,000 genes most strongly interact with the drug. once our screen tells us the MoA, from there we do unbiased screening to identify which parts of which protein(s) in the MoA pathway the drug hits. this puts us in a great position to improve the starting molecule once we know exactly which protein(s) it hits and where. lastly, once we have the new and improved drug and the MoA we can do further unbiased screening to find what disease(s) the new drug works best on
so, MOAT reverses the traditional disease->target->drug process. and because we use unbiased approaches at each step from drug->target->disease, there’s much less build up of sunk costs
for further reading, we have a case study involving a famous class of aging (osteoporosis) drugs called bisphosphonates. PMID: 29745899, 32434850, 30033366.
science.org/doi/10.1126/scit…. see also
@JSheltzer's work where his team tackles the topic in cancer, where the problem is arguably most serious.
science.org/doi/10.1126/scit…
Arash makes several other interesting arguments on the inefficiencies of how drug discovery happens today, such as that most biotech and pharma happens in siloed overly specialized teams, so please check out the article
h/t to
@GolatoTyler for highlighting the original article in
@vitadao’s Discord:
discord.gg/vitadao. side note to all the biotech people out there: there’s lots of alpha like this in various DeSci Discords