Google has funded an artificial intelligence that can accurately predict proteins. I don’t know anything about proteins (I didn’t even know the correct spelling) but I felt you nerds could use this info.
Some of histories most brilliant people did not spell well. Einstein and Churchill to name a few.
5 Likes
About time!!
I played a tiny part in this some 20 years back.
Mainly getting the smart guy pushing the project settled into his new lab and getting his early compute clusters built from salvage while we got the real stuff on order.
The game (CASP) was already afoot 25 years ago…
13 Likes
beautiful stuff. it never ceases to amaze me how all life can be reduced to a simple code, with data flowing one way in a deterministic fashion, providing instructions for the construction of proteins and various sub-types. a base template for all the chemical reactions needed to initiate and maintain living systems.
4 Likes
This is super cool. I’m pretty sure one of my towers still has folding@home running on it in the background.
Didn’t really understand it then, don’t really understand it now. Yay science stuff!
6 Likes
I assume you mean folding of proteins. That’s not too difficult, just computationally intensive.
also requires the training data, which took a wee bit to acquire…
the approach until recently has been to compare an unknown to what is known, and see if maybe there are bits that should fold in a similar manner. not a horrible approach if one assumes all life on the planet is genetically related.
as someone who was assembling compute clusters to perform this sort of work 20 years ago
was an understatement. we’re talking days or weeks of compute time, then more processing to wrap your head around the answer/get it into a format a monkey can grok.
4 Likes
They were doing some of this on one what were termed ‘Beowulf clusters’ when I was in grad school, and preliminarily getting success.
3 Likes
If anyone wants help doing custom proteins/enzymes I’ve been doing a lot of research and design. Also molecular docking of substrate or products. The plan is automate the design process using multiple data sources then eventually test using custom plasmids.
1 Like
Using a method that models drug and target protein interactions using natural language processing techniques, the researchers achieved up to 97% accuracy in identifying promising drug candidates. The results were published recently in the journal Briefings in Bioinformatics.
The technique represents drug–protein interactions through words for each protein binding site and uses deep learning to extract the features that govern the complex interactions between the two.
“With AI becoming more available, this has become something that AI can tackle,” says study co-author Ozlem Garibay, an assistant professor in UCF’s Department of Industrial Engineering and Management Systems. “You can try out so many variations of proteins and drug interactions and find out which are more likely to bind or not.”
The model they’ve developed, known as AttentionSiteDTI, is the first to be interpretable using the language of protein binding sites.
The work is important because it will help drug designers identify critical protein binding sites along with their functional properties, which is key to determining if a drug will be effective.
Sounds like we are quickly approaching the day where we can test drugs in a computer before testing on living beings
In a parallel study:
Accurate and robust prediction of patient-specific responses to a new chemical compound is critical to discover safe and effective therapeutics and select an existing drug for a specific patient. However, it is unethical and infeasible to do early efficacy testing of a drug in humans directly. Cell or tissue models are often used as a surrogate of the human body to evaluate the therapeutic effect of a drug molecule. Unfortunately, the drug effect in a disease model often does not correlate with the drug efficacy and toxicity in human patients. This knowledge gap is a major factor in the high costs and low productivity rates of drug discovery.
2 Likes
…and it gets more complex by getting those instructions “wrong”
1 Like
So how long before we can encode Doom into proteins and make the drug interactions the user input? Cocaine makes you run, meth makes you shoot.