Hi Luca,

This particular error indicates that refinement became unstable for some reason. It is extremely hard to reproduce on another machine, and very well could be that just running the same job on the same machine will go without the error. Nevertheless, we found out that often the reasons for it are:

- using NCS constraints which include only part of a chain, or nonsensical NCS constraints, very different geometry between matching NCS-related parts of the model.
- bad geometry in the starting model, like terrible clashes.

General suggestion to overcome the issue is to check for the above and fix if possible or to increase max_reasonable_bond_distance parameter to e.g. 500.

Please note that excessive_bond_distance_limit is NOT the same parameter and changing it will NOT affect this particular issue.

And for the reference, here is how to change a custom parameter in GUI: 
https://www.youtube.com/watch?v=qB8W_6yuw5k&ab_channel=PhenixTutorials

Or one may choose to supply a parameter in the parameter file or as a command-line argument.

Best regards,
Oleg Sobolev.

On Mon, Sep 5, 2022 at 7:55 AM Luca Pellegrini <lp212@cam.ac.uk> wrote:
Phenix real space refine (dev 4704) crashes with:

RuntimeError : Bond distance > max_reasonable_bond_distance: 67.4676 > 50:

I’ve found various reports of this error in the bb but no (to me) clear solution, only one suggestion to increase the max_reasonable_bond_distance (Nov 2017). I tried setting excessive_bond_distance_limit (I guess is the same) to 999 but with no success as Phenix stops with the same error.

Any help on how to fix this gratefully received.

Best wishes,
Luca






_______________________________________________
phenixbb mailing list
phenixbb@phenix-online.org
http://phenix-online.org/mailman/listinfo/phenixbb
Unsubscribe: phenixbb-leave@phenix-online.org