using identify_cdr_cluster with pyigclassify2

[croshong]

Hi all

I'm now trying to label CDR in my antibody structure using identify_cdr_cluster command in rosetta

I heard that Pyigclassify2 database can be used in rosetta and to use Pyigclassify2 DB

just download pyigclassify2 DB and put all of them in rosetta/main/database/sampling/antibodies.
but even after I download and put pyigclassify2 DB into rosetta/main/database/sampling/antibodies,

identify_cdr_cluster used rosetta/main/database/sampling/antibodies/cluster_center_dihedrals.txt which was used without pyigclassify2 DB

the downloaded pyigclassify2 DB is like below

pyig_ab_mmcif.tar.gz pyig_cluster_mmcif.tar.gz pyig_domain.fasta pyig_readme.txt
pyig_cdr_data.txt.gz pyig_domain_data.txt.gz pyig_mmcif.tar.gz

to use pyigclassify2 in rosetta, should I do some kind of editing of this file ?

or is there any other thing that I miss ?

Thanks in advance

reagards

croshong

[willsheffler]

Thanks for opening this. I’m not sure dropping the PyIgClassify2 download files into is sufficient by itself. From your description, Rosetta is still reading the legacy , which suggests is not automatically consuming those raw archives/tables directly.

My current guess is that Rosetta expects its own preprocessed cluster-center database format, and PyIgClassify2 would need a conversion/update step rather than just copying the downloaded files into place.

A few concrete questions that would help narrow it down:

1. Which Rosetta version/commit are you using?
2. Which exact command line are you running for ?
3. Did you unpack the / files, or are they still in their downloaded compressed form?
4. Are you following any specific Rosetta doc or script for integrating PyIgClassify2, and if so could you link it?

If useful, I can also dig through the Rosetta antibody code/docs to check whether there is a supported import/build step for PyIgClassify2 and report back here.

[roccomoretti]

N.B. Will's message is likely from an AI agent gone rogue. That said, I don't see anything necessarily incorrect in it.