• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

WCG MIP project ending


Senior member
Feb 11, 2008
Per the title, looks like the project is wrapping up as new AI techniques for protein folding are 100s of times faster.

Dear World Community Grid volunteers,

Thank you for your heroic contributions to the Microbiome Immunity Project! Your efforts have allowed us to predict the structures of more than 330,000 proteins contained in the microbiome, dramatically improving our understanding of the functions contained in this incredibly complex community that inhabits all our bodies and underpins many aspects of our health and disease.

With the recent advent of new protein folding techniques based on AI and deep learning, it is now possible to compute these structures hundreds of times faster than when we started the project. We are therefore drawing the project to a close at the end of June, bridging results to new, faster techniques that can be performed on traditional high-performance compute clusters and do not need the power of World Community Grid.

What this means is that your volunteer effort has transformed the accessibility of structural biology for microbiome researchers worldwide, allowing these techniques to be incorporated into the thousands of microbiome projects, targeting diseases ranging from type-1 diabetes and inflammatory bowel disease to cancer to neurological disease, and being conducted around the world.

We already have one scientific article published resulting from the techniques developed from this project, with a second on the way that will apply those protocols to the results that you generated. We will keep you updated on more findings as we publish them.

Thank you again for your impressive contributions to MIP!

Ken g6

Programming Moderator, Elite Member
Dec 11, 1999
Didn't MIP run Rosetta? I wonder if that project will follow? F@H too?


Elite Member
Dec 10, 2016
So-called artificial intelligence, better: machine learning (black box learning) can be applied in fields where there is lots of observational data. Protein structure prediction has become such a field. Yet while MIP (or at least the part of MIP which was put onto the World Community Grid) was about protein structure prediction, F@h's and R@h's focuses are different.

Dr. Bowman of Washington University wrote in December:
Greg Bowman said:
DeepMind’s AlphaFold algorithm has leaped ahead of the pack on the protein structure prediction problem. I won’t go into the technical details, but will just say congrats to the team for the big step forward.

With that said, AlphaFold doesn’t explain how proteins fold, which is another important piece of the protein folding problem. It also doesn’t solve the host of other problems that are closely linked, if not part of, the protein folding problem.

Folding@home’s original focus was on understanding how proteins fold into their dominant structures. Since then, the project has broadened its scope to encompass many related problems, since all of these phenomena are driven by the same underlying physical principles.

Here is an example article which highlights that protein–protein interaction is of dynamic, not static, nature: source, also by Dr. Bowman in December 2020.

Similarly, protein structure prediction is but one part of the research which the scientists who are associated with the Rosetta@home project are undertaking, but their main focus is on protein design nowadays.

Interestingly, they did publish a paper in January 2021 about the involvement of deep learning tools in the analysis of observational data in protein dynamics: Ziatdinov (ORNL) et al, "Quantifying the Dynamics of Protein Self-Organization Using Deep Learning Analysis of Atomic Force Microscopy Data" (Bakerlab publications, article). From what I understand from the abstract, 2-dimensional protein unmixing experiments which are slow enough to be observed through a microscope, are used as observational data in a deep learning workflow, which then helps classify state transitions in these particle systems, which then is meant to help in deriving the fundamental principles behind these transitions.
Last edited: