Starting to get a little stir crazy so it’s time for the unpopular opinions game. Gimme a topic, I’ll give you an opinion.
-
-
Replying to @s_r_constantin
one legitimately exciting use for maching learning, and one overhyped stupid use please
1 reply 0 retweets 4 likes -
Replying to @gravity_levity
exciting: vaccine optimization. overhyped: QSAR.
1 reply 0 retweets 13 likes -
Replying to @s_r_constantin @gravity_levity
what's a good resource on the former?
1 reply 0 retweets 3 likes -
Replying to @jsdenain @gravity_levity
what little I know I learned from Alex Rubinsteyn http://www.rubinsteyn.com/ and keeping up with new biotech startup news
1 reply 0 retweets 2 likes -
also rough intuition: the number and distribution of cell surface proteins & antibodies is in the range where ML seems like a big improvement (huge diversity between individuals, combinatorial explosion making traditional stats impractical)
2 replies 0 retweets 2 likes -
and you can have a tight in-silico/in-vitro feedback loop for designing/optimizing this stuff
2 replies 0 retweets 2 likes -
computing binding affinities of proteins and small molecules has a.) a 50-year history of abject failure; b.) is easily tested experimentally; c.) is sparse as hell so maybe you don't need ML as much; d.) mOsT aPpRoVeD dRuGs dOn'T hAvE tHeIr pUtAtIvE mEcHaNiSmS oF aCtIon
3 replies 0 retweets 7 likes -
Seeing all these coronavirus drug repurposing screening papers with notes that they don't even know whether the drug-protein interaction will inhibit the target protein or enhance the stability of its interactions.
1 reply 0 retweets 3 likes -
Replying to @__ice9 @s_r_constantin and
And also suggesting things like emetine and chlorpromazine, but that's beside the point.
2 replies 0 retweets 2 likes
aaaaaaaaaaaaaaa
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.