Read on Twitter
Had an amazing call @thebiodojo with @rbhar90. We talked about open source code versus true biological insights, the variety of skills a founder needs in a biotech startup, and how IMPORTANT data streams are > model architecture.
Here are my key takeaways
Here are my key takeaways
1/ Biotech founders should be able to communicate more than the computational aspects these days. Before, you could get funding for simply saying drug discovery x AI.
Computational stack is easy to open source, biology is not as easy to open source. Know your basic biology WELL.
Computational stack is easy to open source, biology is not as easy to open source. Know your basic biology WELL.
2/ When you are building a startup, it’s 5% heads and 95% tails. We convince ourselves it’s 50-50.
INVEST TIME. Build your network of biotech founders and mentors. Understand VC cycles well.
INVEST TIME. Build your network of biotech founders and mentors. Understand VC cycles well.
3/ If you’re looking to breakthrough in deep science. Focus is essential.
However scientific founders need to be the jack of all trades. Know operations, sales, hiring, and dealing with random bullshit that gets thrown at you.
However scientific founders need to be the jack of all trades. Know operations, sales, hiring, and dealing with random bullshit that gets thrown at you.
4/ Bringing computational chemistry into clinics require lots of DATA.
We need more assays to allow for larger data streams like DNA encoded libraries or assays specifically for protein design. Better microscopy systems.
We need more assays to allow for larger data streams like DNA encoded libraries or assays specifically for protein design. Better microscopy systems.
5/ The hard truth of #AIinMedicine is that only 10% of biological is affected by machine learning.
Areas like developmental biology are hard to apply ML. #COVID19 is an example where AI is hard for small molecule design since the basic biology of the virus is still unknown.
Areas like developmental biology are hard to apply ML. #COVID19 is an example where AI is hard for small molecule design since the basic biology of the virus is still unknown.
