I have a peptide X of length 50aa. Alphafold2 predicted the structure which takes a helix form and with high average pLDDT score (>90).
But during actual experiment (e.g. crystallography, spectroscopy), peptide X is determined to have a random coil structure.
My question is how can I resolve the difference between AF2 prediction and crystallography?
To what extend we can use AF2 peptide structure prediction for downstream analysis (e.g. docking, etc)?
I find it hard to believe that a peptide that has only a random-coil structure was resolved by X-ray crystallography. By definition, those are too flexible to crystallize in a single conformation. Are you thinking that crystallography solved your peptide as a coiled-coil? If so, that's a long alpha helix, in which case the AlphaFold2 prediction could be accurate.
It could be that your peptide has an alpha helix structure in a protein context, when not completely exposed to solvent, and a random coil structure in isolation when it is completely solvent-exposed.
Not just in your case, but in general: I don't think one should ever use the AlphaFold2 prediction if a crystal structure is available. It is another question altogether how you were able to crystallize and solve the structure of a random coil peptide.
Mensur Dlakic Thanks for your comment. I was just told that the peptide X is a random coil, by the experimenter. They don't have the actual structure (e.g. PDB file) of the peptide. I think you're right when you said peptide X has "random coil structure in isolation when it is completely solvent-exposed". I just wonder if that is the case, can we still reliably use AlphaFold2 structure of peptide X for downstream analysis?
It doesn't seem like you have a choice if no PDB file is available for the real structure. If you are lucky, your peptide gets ordered into a helix upon binding its partner, in which case the AlphaFold2 model may end up being accurate. Either way, I don't think anyone can tell you with certainty that using this model is a good idea or not. I suggest you give it a try and see what happens. I am assuming that you are using flexible docking where both the ligand and the target are allowed to change conformations locally.
Mensur Dlakic Thanks for your comment. I was just told that the peptide X is a random coil, by the experimenter. They don't have the actual structure (e.g. PDB file) of the peptide. I think you're right when you said peptide X has "random coil structure in isolation when it is completely solvent-exposed". I just wonder if that is the case, can we still reliably use AlphaFold2 structure of peptide X for downstream analysis?
It doesn't seem like you have a choice if no PDB file is available for the real structure. If you are lucky, your peptide gets ordered into a helix upon binding its partner, in which case the AlphaFold2 model may end up being accurate. Either way, I don't think anyone can tell you with certainty that using this model is a good idea or not. I suggest you give it a try and see what happens. I am assuming that you are using flexible docking where both the ligand and the target are allowed to change conformations locally.