Ignore certain PDBs during the building process

[joaomcteixeira]

Some users are asking how not to consider certain PDBs' torsions when creating conformers. This is relevant when building proteins (or parts of proteins) already deposited in the PDB. It is a very good question we have discussed many times. There are two ways to achieve this:

1. Unlist those PDBs from the initial PDB culled list when creating the database, or
2. Remove those entries from the JSON final database file.

We need two things:

• A tutorial page explaining this issue and how to address it
• A new client helping people to remove specific PDB entries from an already created JSON torsion database or a flag in build to ignore certain PDBs from the building process (likely this one is better).

For example, to ignore PDBs 1ABC and 2DEF from the torsion sampling:

idpconfgen build [...] --ignore 1ABC 2DEF

[menoliu]

Good idea! I think the search sub-client would be handy to help solve this issue.

[joaomcteixeira]

I think search is a very broad word. Why another subclient? To edit the database? It can be. Can we forge some examples to prototype the interface before starting coding?

Cheers! 🚀

[menoliu]

Oh I meant for the tutorial page, users can use the existing cli_search to find which PDB IDs they need in the database already, and during the build step, they can specify which PDB IDs to ignore from the database like: -ign 2MX4 to ignore all chains/segements of 2MX4

[joaomcteixeira]

Hi Nemo,
-ign sounds good. We could also allow -ign to accept PDBIDs not in the database. In this way, we (or the user) could have provide (have) "the" database and ignore at will.