Saturday, March 30, 2019

MALDI-TOF + Machine learning = New way of finding drug resistance mechanisms??


How powerful is machine learning getting? For a distraction check out QuickDraw with Google. You draw stuff and a neural network thingy can even guess some of the stuff that I draw! (Those are mine above). It didn't get "beard" or "phone". It occurs to me that I forgot what a phone looks like.

For an example more relevant to the title of this blog -- check out this new preprint!


What did they do with machine learning? They learned biological stuff from bacteria and MALDI-TOF -- stuff about mechanism of action of different drugs on those bacteria!

Am I qualified to review this in any way? Nope. I flipped through the words, saw a lot of math and absorbed none of it. I can tell you they used a Bruker MicroFlex LT and only scanned from 2,000 to 15,000 m/z -- which seems like a smallish mass range for non-digested bacterial cells blasted with a laser -- but appears to provide sufficiently differentiating data. They also used some interesting programs in MatLab that I hadn't heard of for some of the pre-machine learning stuff.

In the end, I think it's a great Saturday afternoon read on something completely different than what I am supposed to be doing!

No comments:

Post a Comment