Google’s deep-learning program (AlphaFold) for determining the 3D shapes of proteins stands claims that the AlphaFold, outperformed around 100 other teams in a biennial protein-structure prediction challenge called CASP.
Dear M.M. Mohamed Mufassirin ,
I recommend a well written and nuanced description how to estimate the true value (and ethics) when it comes to AlphaFold: https://dasher.wustl.edu/bio5357/discussion/oxford-alphafold2.pdf One important conclusion I think is important (and I think I agree) is: that it might be that AlphaFold 2 has solved the problem of protein prediction but certainly not “as many press releases have claimed, the protein folding problem. DeepMind’s code will provide no information of how a polypeptide, or an ensemble of chains, assembles within seconds into the intricate structure it requires to function. It can just provide an accurate estimation of the crystal structure, which is just a snapshot of the conformation of the protein.”
Best regards.
Dear M.M. Mohamed Mufassirin ,
I recommend a well written and nuanced description how to estimate the true value (and ethics) when it comes to AlphaFold: https://dasher.wustl.edu/bio5357/discussion/oxford-alphafold2.pdf One important conclusion I think is important (and I think I agree) is: that it might be that AlphaFold 2 has solved the problem of protein prediction but certainly not “as many press releases have claimed, the protein folding problem. DeepMind’s code will provide no information of how a polypeptide, or an ensemble of chains, assembles within seconds into the intricate structure it requires to function. It can just provide an accurate estimation of the crystal structure, which is just a snapshot of the conformation of the protein.”
Best regards.
Dear Rob Keller ,
Thank you very much for your opinions and sharing. Really it is helpful. Much appreciated.
Regards
Mufassirin
It will be surely helpful in applied science tasks like drug discovery or simulation of mutation , it is not a breakthru as for basic science where problems like allosteric effect, molecular machines, creation of liquid-liquid interfaces ask for physically relevant models and not brute force.
- Badges
- Science topic