Simple IgBLAST wrapper

[Jason Vander Heiden]

Now that we have all the IMGT-gapping, indel, junction and FWR/CDR issues resolved with MakeDb-igblast, it might be worth it to write a very simple IgBLAST wrapper that just calls IgBLAST with the optional arguments we require. I'm not interested in duplicating migmap or doing anything that would be complicated to maintain - just something that builds the db, if needed, and calls IgBLAST. I think we should still output the IgBLAST naive fmt7 files for import into MakeDb, as this fits better with our modular approach to design.

[Jason Vander Heiden]

We can have the same tool wrap blastn and usearch, and use it for arbitration searches, such as filtering out Phi-X reads.

[Jason Vander Heiden]

This shell script works file, so we can just port it to python and add it to changeo formally: https://bitbucket.org/kleinstein/docker/src/tip/immcantation-base/tools/run_igblast.sh

[Roy Jiang]

will probably take like 2 weeks for me though.

[Jason Vander Heiden]

Partially finished as AssignGenes-igblast (name subject to change). Still need to add a separate subcommand to build the igblast database, which just involves porting imgt2igblast.sh.

[Jason Vander Heiden]

Done in d7c203f. I don't think we need to add a database building wrapper. It's a bit too error prone an operation.

[Roy Jiang]

Would it better to build the database during the setup.py install step?

[Jason Vander Heiden]

No, I think that'd be too error prone. And I think it's out of scope anyway.

We don't need to get into doing the igblast install (outside the docker container) and setting up the database is part of the igblast install.

[Jason Vander Heiden]

Bah. I spoke too soon. There is no default database for humans, so you have to build one. Let me re-examine the options.

[Scott Christley]

This may take longer than you like, but AIRR germline WG is formulating a set of standard (human, mice of various strains, and macaque are planned) germline sets with annotations. I've already gotten verbal approval from Jian @ NCBI that they will change IgBlast to use those annotations files instead of the internal_data that they currently have. It's going to be simple step to write a script that takes those germline sets and generate IgBlast databases for them. That actually might reside in airr-standards along with an API for tools to read and use the germline annotations.

[Jason Vander Heiden]

Thanks @schristley, that does help. I'm going to chew on it for a couple days, but I'm really leaning against the idea of adding igblast installation steps to what should be a simple wrapper. And if it's going to be obsolete in the near future... meh.

[Jason Vander Heiden]

Closing. Still not decided, but there realistically isn't time to resolve this right now. We can revisit in the future if it becomes an issue.