I have been reading the following article on peptide structure prediction. https://www.biorxiv.org/content/10.1101/2022.02.17.480937v1.abstract I wanted to know whether alpha fold 2 can be used to generate distributions of structure of a given peptide sequence. In other words, Can one generate distributions of backbone Ramachandran angles of a given peptide sequence using structures predicted from various conformations generated by alpha fold2?
If you use alpha fold to generate peptide structures, they may not be significantly accurate. One of the solution is - try generate the peptide structure using multiple tools capable of producing secondary structures (including alpha fold) and go for molecular dynamics simulation, run DSSP analysis and compute/model the average structure.
Kangkon Saikia Thanks for answering ! Why alpha fold is inaccurate ? Is it possible to construct probability distribution of backbone Ramachandran angles of a given peptides sequence using alpha fold2 ?
There are many articles and news published about various shortcomings of alpha fold. Moreover, alpha fold is like a generative AI which highly rely on MSA with template peptides. For those sequences having a proper identity to a certain template can only be heuristically predicted with alpha fold. As an example of a problem, it does not inculude any ab initio algorithms for predictions of secondary structures in a region which it has never seen in the training dataset.
And for construction of probability distribution of backbone Ramachandran angles, I am unaware of any alpha fold feature to do that. The AI model is built in such a way that it would produce 3D coordinates of the protein structure with a score of their prediction accuracy.
Kangkon Saikia Could you suggest some more reference in this directions where the failure of alpha fold is pointed out ?
you can follow these article and cross references cited.
Article AI revolutions in biology: The joys and perils of AlphaFold
https://www.chemistryworld.com/opinion/why-alphafold-wont-revolutionise-drug-discovery/4016051.article
- Badges
- Science topic