Why I left an academic career to build ... mouse tech?

Animal research is not a sexy space in biology. It's repetitive, dirty and runs on 1980s tech. It doesn't get you publications in high-impact journals.

I left a fairly successful academic career built over 15 years as a genetic engineer to improve animal research. Why?

Because currently:

-90.4% of drugs fail in humans.

-It costs $2.3B and takes 12 years to get a drug to market.

-Only ~50 new drugs are approved per year in the entirety of the US.

-A basic mouse longevity study costs half a million when outsourced.

The numbers above are largely a phenotype of the state of the art in animal research.

It doesn't matter what interventions you can engineer if you can't test them effectively. And animal studies are absolutely vital - there is no way around them. This is the bottleneck that hit me head on, full force, in my PhD.

In my PhD, I ran 4 studies containing a total of 500 animals to test longevity gene therapies. Testing means measuring one animal at a time, for one aspect of health, by hand, replicating tens of times, and writing the data in a paper notebook. This is expensive, exhausting and not reproducible. You need a team of well trained technicians to generate trustworthy data - or run it all with your own hands. The latter is what I did - and it took me 8 full months of doing solely manual measurements all day, every day to get a basic dataset.

Not only is the state of handling and measuring animals poor - but animal models themselves are often chosen for convenience and habit rather than proper humanization and maximum predictive validity (a topic for a different post).

As a consequence, few quality studies are published each year. Most are underpowered. Very few are ever replicated.

Yet a study in a mammal (usually mouse) is the data that people rely on to consider something potentially effective.

This is the key bottleneck. By improving predictive validity of animal studies, you can have a massive impact downstream. By improving throughput and cost, you can test more interventions and select the actual winners.

By basic math, if as a result of the above, you can reduce the failure rate by 2% (from 90% to 88%), 10 more drugs would be approved per year in the US - more than the output of the three biggest pharma companies combined.

This is the power of targeting a key bottleneck.

There is so much to be done in animal research. Automation, cost reduction, reproducibility, humanization, standardization. True - animals will never be perfect models for humans - but today they are very far from their potential.

This is why I went into animal research.

Many more people should as well.

Original post here (for comments)