Clustering failure - IMPROPER tag in topology

HADDOCK
1

Hello, I am running into some issues with the clustering of my HADDOCK3 run where clusters are forming, but they comprise of exactly four identical models.

Additionally, I am finding some atoms under the IMPROPER tag within the topology file of the ligand, and I can see that they are in a tetrahedral configuration around a central carbon. I suspect that perhaps the IMPROPER tag is flattening this shape into a plane and causing the CNS engine to halt the simulation to prevent the run from crashing. I have highlighted it in yellow; it is the first group which appears under the tag in the .psf file.

1UU6_1OLR_2_3_18_5
1UU6_1OLR_2_3_18_53279×1650 217 KB

I have a suspicion that maybe if the CNS engine halts the simulation due to the energetic issues with the IMPROPER tag, that perhaps the models in the clusters are somehow generated from the [rigidbody] results for the [flexref] module - which then crashes - and leaves us with multiple identical models.

I have attached below the run directory and configuration file. Any help would be greatly appeciated!

2

Your run completed fine.

And improper are used also to maintain chirality for example.
So it is normal to have one defined for the carbon you are highlighting. And actually crucial for glycans

Your workflow is also a bit strange. Why do rigidbody if your are randomising the starting orientations in flexref?

Consider removing from your flexref module the following parameters:

sampling_factor = 10
randrot = true
separate = true

Or: remove the rigidbody module and increase the sampling_factor to have the number of models you want to generate.
This would be equivalent to the protein-glycan example: docking-flexref-protein-glycan-full.cfg

Now you have a mixture of two different workflows that does not make sense.