CDD database

[fmaumusINRA]

Hello,

Doing some protein work, among the databases proposed for download, would it be possible to add a profile library corresponding to the CDD database? I need to search for protein domains.

Thanks,
Florian

[milot-mirdita]

This should work:

wget https://ftp.ncbi.nih.gov/pub/mmdb/cdd/fasta.tar.gz
mmseqs tar2db fasta.tar.gz cddmsa
sed 's|\.FASTA||g' cddmsa.lookup > cddmsa.lookup_tmp
mv -f cddmsa.lookup_tmp cddmsa.lookup
mmseqs apply cddmsa cddmsa_wo_cons -- awk '/^>/ { i++ } i > 1 { print; }'
awk 'BEGIN { printf("%c%c%c%c",11,0,0,0); exit; }' > cddmsa_wo_cons.dbtype
ln -s cddmsa.lookup cddmsa_wo_cons.lookup
mmseqs msa2profile cddmsa_wo_cons cdd --match-mode 1

mmseqs easy-search QUERY.fasta cdd res.m8 tmp

Could you try out if this performs approximately how you expect it to? If this works well I can add it to the databases workflow.

(I fixed a few things shortly after I posted this.)

[fmaumusINRA]

It works like a charm, thanks a lot.
I think it is a very useful database and, I don't know why, I can find pfam domains that were missed when using the pfam database (both seed & full).

[milot-mirdita]

Could you share one query that finds a PFAM match in CDD and not in PFAM(-full/seed)?

The PFAM databases are essentially built with the same procedure as described above.

[fmaumusINRA]

Sorry I think that was was the case when using pfam as querydb, experimenting...
Anyway, thanks a lot for providing CDD database!!

[milot-mirdita]

I added the CDD to the databases downloader. Thanks for the suggestion!