FCC with custom selection

I am trying to run FCC clustering ignoring custom chains. I do not have a symmetric oligomer. Instead, I have subunits that are placed in an EM map and therefore fixed during the run, combined with mobile subunits.
I guess I can play with the FCC threshold so that my cutoffs are more stringent and the contacts are not “diluted” by all the common contacts between fixed subunits, but I was wondering if there is somewhere I can insert a chain/segid selection?
thank you!

Hi Andrea,

Not out of the box, unfortunately. Selection was something I wanted to add for a while but never had the time to.

What you can do - I’ve done it myself - is to run the first step (make_contacts) and then grep/sed the contacts you want to a new set of files and use those to run the last two steps (calc_fcc_matrix and cluster_fcc).

Let me know if you have any ideas/issues.



A terça, 9/02/2021, 06:03, andreagraziadei via BioExcel <bioexcel@discoursemail.com> escreveu:

Thanks for the prompt response. I will do grep/sed. Unfortunately, I don’t know any cpp, so I have no idea where to look there… I know that one can do this in Bio.PDB relatively easily with NeighborSearch, so I could reimplement it there, but I bet you know a lot more about this than me :wink: Anyways, this solution is definitely good for now!

Ah, I meant the output “.contact” files that should appear in the analysis/ folders. These are text files, one per model, that have one residue-residue contact per line in the format: resAsegAresBsegB, where segA is always the 6th character and segB the last one. So if you want to exclude certain chains, you need to filter whatever segs you want. The chains will be converted to a number, sequentially by order of appearance in the file, so if you have say, chain B chain C chain A, the mapping would be B:0, C:1, A:2.

Actually, reading the code for FCC (the standalone version, not the scripts packaged with HADDOCK), there is already a selection option in there :slight_smile: It’s in make_contacts.py, the -s flag. I don’t remember writing it though… so I am not sure it’s super tested :slight_smile: You could get the code here (https://github.com/haddocking/fcc) and use those scripts instead.

andreagraziadei via BioExcel <bioexcel@discoursemail.com> escreveu no dia quarta, 10/02/2021 à(s) 06:04:

The other solution would be to filter out the chains you want to exclude from your clustering by creating new PDB files with our pdb-tools scripts