Spaces:
Running
Drug Discovery & Scientific Research
Collectively, we could all make a huge difference to scientific research and drug discovery if we combined our daily passion for trying out new models with tools that can utilise our local hardware.
I'd love to discuss how this could be achieved with the scientific and medical communities for a global shared tool (or service), that may work on a cluster basis for example, utilising the tools and models that have been developed or are being developed for scientific discovery and research.
This thread maybe of interest for @cyrilzakka 🔥
hello benjy open/acc :-) :-) let's make something real happen in this space 🚀
These models are a good place to start: https://huggingface.co./models?other=biology & https://huggingface.co./models?other=chemistry
The recent open source release of Alpha Fold 3 weights and Alpha Proteo could be used in a pathway for drug discovery. There would be other parts of the system needed, but those two do the bulk of the drug development work on the protein and protein interactions level.
I totally don't underestimate the challenge, effectively what we are asking is that someone with experience in the field line up the tasks (which ideally are not accessible params for those outside the field to tamper with), and then when the scientists / researchers are ready we all jump on board with the network cluster to provide the collaborative compute. I'm thinking Exo Labs might be able to help here, or sites like this seem like they are on the right track:https://www.primeintellect.ai/
The only issue with this: Sites like this that spawn clusters on a warehouse basis appear to be very disconnected from both end users: the researchers and the providers.
If people knew about what cause they were providing their GPUs for, instead of just chasing a quick buck, I like to believe there would be a good group of individuals motivated to get involved.
My colleague at Copyleft Cultivars, teaching at Portland State University, has a grant application project pending which, if fruitful, may be interested in using open source ML pipelines and community collaboration here for using drug discovery AI systems in the context of regenerative farming practices/treatments. Finding novel treatments for plant diseases such as septoria, could be another unifying cause among a grassroots community, because we all eat and rely on food from plants battling these diseases. Those may be good test cases for the type of projects we could help.
Plant disease treatment is much easier to test and verify the pipeline and results IRL, vs human or animal research.