To answer your question as best I can: it's exploratory at the moment, like the examples in the original article. The immediate goal is to pursue directions one of which will hopefully produce software and knowledge clearly valuable enough to warrant significantly increasing resources devoted to it--the next inflection point. Risks and rewards are high at this stage. I won't go into a why starting or joining a startup is fantastic, though, as PG does that just fine[1].
Capturing created value outside of the research arm of a big pharma company is of course a key challenge in bioinformatics, and I'd love to meet or chat with you if you're thinking about this too.
In the missionary vs mercenary divide that John Doerr uses to segment startups, I'm now a missionary. The end goal is to cure disease.
I don't need convicing that it is easy to inovate outside of big pharma. Altough bioinformatics startups do require quite a bit more cash then pure software startups.
However, the biggest part of capturing value is the regulatory burden.
In other words, lets say you have a great way to discover biomarkers. Even more so, you go ahead and DO discover a pile of biomarkers with great potential.
Then you attemp to sell those biomarkers to big pharma, and that is when you find out:
Even with all their beurocracy and horrible return on R&D investments, big pharma is sitting on a huge pile of biomarkers.
Pushing any one tratment through the pipeline costs a metric tonne of cash and takes 10 year before you even know if it's worth to keep going.
That is why there's a huge backlog of potential drugs, biomarkers, treatments, etc.
Thus merely finding new things with potential is only slighlty better then nothing.
Capturing created value outside of the research arm of a big pharma company is of course a key challenge in bioinformatics, and I'd love to meet or chat with you if you're thinking about this too.
In the missionary vs mercenary divide that John Doerr uses to segment startups, I'm now a missionary. The end goal is to cure disease.
[1]: http://paulgraham.com/notnot.html; http://paulgraham.com/boss.html; http://paulgraham.com/articles.html