ColabFold
ColabFold copied to clipboard
sokrypton
Aberrant 3D modeling of a multiprotein complex leads to the formation of a covalent interprotein bond Arginine - Leucine
I attempted to model a missense mutation in the NUF2 peptide located in the NDC80 protein complex. The substitution of a leucine at position 303 by an arginine in the NUF2 peptide leads to a covalent interprotein bond with the leucine opposite on the NDC80 protein at position 463. This interaction would therefore reinforce the quaternary structure of this complex. Unless I am mistaken, this type of interaction cannot take place and I expected rather a destructuring of the region.
On the attached figure showing the modeling leading to the covalent bond between NUF2 and NDC80, NUF2 appears in green and NDC80 in yellow.
Could you explain to me where this issue comes from? Are there any parameters to change?
AlphaFold2.ipynb
{
"num_queries": 1,
"use_templates": false,
"use_amber": false,
"msa_mode": "MMseqs2 (UniRef+Environmental)",
"model_type": "AlphaFold2-multimer-v2",
"num_models": 5,
"num_recycles": 3,
"model_order": [
1,
2,
3,
4,
5
],
"keep_existing_results": false,
"rank_by": "multimer",
"pair_mode": "unpaired+paired",
"host_url": "https://api.colabfold.com",
"stop_at_score": 100.0,
"recompile_padding": 1.0,
"recompile_all_models": false,
"commit": "1f2d963e7705b0d8c847980e4b1c44243127febb",
"version": "1.2.0"
}
AlphaFold2 as well as AlphaFold-multimer were trained to be robust against mutations. Both models are not the right tool to model single point mutation effects.
The bond you see is just clashing. If you use amber to relax the clashes, then this artifact will go away. That said AF2 is not sensitive to single point mutations.
Thank you very much for your answers. I will redo the analysis using Amber. What model would you suggest to use to study the impact of a single point mutation?
That's still a hard problem, but maybe something like: https://pubs.acs.org/doi/10.1021/acs.jpcb.7b11367