What makes this dataset or problem worth solving compared to other health datasets? Would the results on this task be broadly useful to health?
CyberDildonics•Apr 4, 2026
What other "datasets" are you talking about? How do you "solve a dataset" ?
pfisherman•Apr 4, 2026
Nice work! Here is an article you may find helpful if you have not already come across it.[0]. You may also want to consider benchmarking against some non ML methods.[1]
It’s a self supervised learning architecture, and it’s pretty much universal. The loss function runs on embeddings, and some other smart architectural choices allover. Worth diving into for a few hours, Yann LeCun gives some interesting talks about it
simianwords•Apr 4, 2026
What makes these Domain specific models work when we don’t have good domain models for health care, chemistry, economics and so on
colechristensen•Apr 4, 2026
>we don’t have good domain models for health care, chemistry, economics and so on
Who says we don't?
simianwords•Apr 4, 2026
Examples please?
colechristensen•Apr 4, 2026
No, it's really simple to search for domain specific models being used "in production" all over the place
simianwords•Apr 4, 2026
I didn’t find a single one that outperforms a general model.
colechristensen•Apr 4, 2026
Ok, alphafold.
simianwords•Apr 4, 2026
It’s not a large language model
rubicon33•Apr 4, 2026
Can someone explain what one might use this model for? As a developer with a casual interest in biology it would be fun to play with but honestly not sure what I would do
colechristensen•Apr 4, 2026
You can get your feet wet with genetic engineering for surprisingly little money.
Lab strains of things tend to be extremely sensitive and not human adapted. You shouldn't study and modify human-infecting organisms in your basement anyway. While you shouldn't ignore protective equipment and proper procedure... paranoia about infecting yourself with a lab leak isn't warranted.
yieldcrv•Apr 4, 2026
Distributing the load on this will probably be infinitely more useful than “folding at home”
skyskys•Apr 4, 2026
hmmmm seems like some fake hype.
seamossfet•Apr 4, 2026
The problem with models like this is they're built on very little actual training data we can trace back to verifiable protein data. The protein data back, and other sources of training data for stuff like this, has a lot of broken structures in them and "creative liberties" taken to infer a structure from instrument data. It's a very complex process that leaves a lot for interpretation.
On top of that, we don't have a clear understanding on how certain positions (conformations) of a structure affect underlying biological mechanisms.
Yes, these models can predict surprisingly accurate structures and sequences. Do we know if these outputs are biologically useful? Not quite.
This technology is amazing, don't get me wrong, but to the average person they might see this and wonder why we can't go full futurism and solve every pathology with models like these.
We've come a long way, but there's still a very very long way to go.
stardust2•Apr 4, 2026
How do we get more verifiable protein data? So even if we had better data, we don't yet understand how the structure impacts the biology?
colingauvin•Apr 4, 2026
HN's blindspots never cease to amaze me.
I am a structural biologist working in pharmaceutical design and this type of thing could be wildly useful (if it works).
9 Comments
0. https://pubmed.ncbi.nlm.nih.gov/35318324/
1. https://www.nature.com/articles/s41586-023-06127-z
JEPA is going to break the whole industry :D
Who says we don't?
This guy shows a lot of how it's done: https://www.youtube.com/@thethoughtemporium
Basically you can design/edit/inject custom genes into things and see real results spending on the scale of $100-$1000.
Lab strains of things tend to be extremely sensitive and not human adapted. You shouldn't study and modify human-infecting organisms in your basement anyway. While you shouldn't ignore protective equipment and proper procedure... paranoia about infecting yourself with a lab leak isn't warranted.
On top of that, we don't have a clear understanding on how certain positions (conformations) of a structure affect underlying biological mechanisms.
Yes, these models can predict surprisingly accurate structures and sequences. Do we know if these outputs are biologically useful? Not quite.
This technology is amazing, don't get me wrong, but to the average person they might see this and wonder why we can't go full futurism and solve every pathology with models like these.
We've come a long way, but there's still a very very long way to go.
I am a structural biologist working in pharmaceutical design and this type of thing could be wildly useful (if it works).