Forums  > Off-Topic  > AI to find new drugs  
     
Page 1 of 1
Display using:  

Jurassic


Total Posts: 373
Joined: Mar 2018
 
Posted: 2020-04-01 08:50
Does anyone know where I can read more about using AI to scan for new drugs?

(Real science not popular science)

mtsm


Total Posts: 253
Joined: Dec 2010
 
Posted: 2020-04-02 02:41
Do you mean chemometrics?

scurve


Total Posts: 18
Joined: Aug 2009
 
Posted: 2020-04-02 03:16
There are so many journals and start-ups. Check out https://benevolent.ai/publications, these folks are legit.

Jurassic


Total Posts: 373
Joined: Mar 2018
 
Posted: 2020-04-02 13:57
thanks guys, thats a good start

hftpm


Total Posts: 4
Joined: Oct 2013
 
Posted: 2020-04-03 18:11
https://www.deshawresearch.com/

Maggette


Total Posts: 1257
Joined: Jun 2007
 
Posted: 2020-04-03 18:58
More bio and medical research in general, but maybe something for you..the guys and girls at google:).

https://research.google/pubs/?area=health-bioscience

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

rod


Total Posts: 413
Joined: Nov 2006
 
Posted: 2020-04-03 20:07
IBM RXN: "the first, free AI web service for predicting chemical reactions."

Seems to be the result of a research project that produced papers like this.

Jurassic


Total Posts: 373
Joined: Mar 2018
 
Posted: 2020-04-13 20:33
@mtsm I tried chemometrics textbooks on Amazon and they were indistinguishable from statistics textbooks

mtsm


Total Posts: 253
Joined: Dec 2010
 
Posted: 2020-04-14 01:44
I wasn't suggesting you look at textbooks on amazon. I was asking you if by the wording of your question about "using AI to scan for new drugs" you meant chemometrics.

Not too surprised about your findings, although at some level I would expect that would find some domain knowledge weaved into the statistics.

phopstar


Total Posts: 9
Joined: Jul 2020
 
Posted: 2020-07-20 22:10
This is in excellent paper: https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.7b01144

A key concept to grasp when evaluating this stuff is understanding how severely vast chemical space is. Sampling a small corner of it and then using that as a training set will not get you a model that generalizes well unfortunately.
Previous Thread :: Next Thread 
Page 1 of 1