hmmscan in TEsorter need hmmpress first

[baozg]

Hi, @oushujun

When using cds to clean up the TElib.fa, the TEsorter may throw the following error (have seen similiar error in other issues #105 )

b'\nError: Failed to open binary auxfiles for /data/software/conda_envs/EDTA/bin/python3.6/site-packages/TEsorter/database/REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm: use hmmpress first\n\n'

It need the first run the hmmpress for the hmm you use, run the following command:

# your conda envs
cd /data/software/conda_envs/EDTA/lib/python3.6/site-packages/TEsorter/database
for i in `ls *hmm | sed 's/.hmm//g'`; do hmmpress ${i}.hmm > ../${i}.hmm;done

[oushujun]

@Juke34 @zhangrengang Is it possible to have this step done automatically during installation or the first-time use?

[Juke34]

Yes it can be done in the conda recipe

[Juke34]

It sounds @zhangrengang added this step in newer version of TEsorter.
See log when installing v1.3

SERR 2020-10-29 22:26:30,093 -INFO- check database rexdb
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR 2020-10-29 22:26:30,094 -INFO- db path: /usr/local/lib/python3.9/site-packages/TEsorter/database
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR 2020-10-29 22:26:30,094 -INFO- db file: REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR 2020-10-29 22:26:30,094 -INFO- db REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm not yet ready, building db!
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR 2020-10-29 22:26:30,571 -INFO- Working...    done.
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR Pressed and indexed 233 HMMs (233 names).
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR Models pressed into binary file:   /usr/local/lib/python3.9/site-packages/TEsorter/database/REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm.h3m
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR SSI index for binary model file:   /usr/local/lib/python3.9/site-packages/TEsorter/database/REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm.h3i
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR Profiles (MSV part) pressed into:  /usr/local/lib/python3.9/site-packages/TEsorter/database/REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm.h3f
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR Profiles (remainder) pressed into: /usr/local/lib/python3.9/site-packages/TEsorter/database/REXdb_protein_database_viridiplantae_v3.0_plus_metazoa_v3.hmm.h3p
22:26:30 BIOCONDA INFO (ERR) [Oct 29 22:26:30] SERR

[Juke34]

Apparently it was also present in previous version.
@baozg How did you install EDTA?
What are the commands to end up in this state? (i.e. what do you mean by using cds to clean up the TElib.fa).

[zhangrengang]

Hi @Juke34. Can hmmpress or build_database.sh be run when installing but not when running at the first time? There is an issue that non-root users have no permission to build the database when running TEsorter (throwing Error: failed to open SSI index *.hmm.h3i when hmmpress and Error: Failed to open binary auxfiles for *.hmm: use hmmpress first when hmmscan) if it was installed by root.
In this way, the issue #121 (maybe also caused by permission) should be also solved as all database have been present.

[Juke34]

Ok I understand now why it is important to do so at the installation. Ok I will try to add it.

[oushujun]

@Juke34 Any luck with adding hmmpress to the conda recipe?

[Juke34]

Yes I did it last month, see above Tesorter bioconda/bioconda-recipes#25148. Merged

[oushujun]

I see. The Biocontainer's docker version of EDTA mentioned in #122 still has the hmmpress issue. I just check the TEsorter version in this docker image, it's 1.2.5.2. Could it be that this image was created before TEsorter got updated and even I freshly pull it from the hub it is still using the old version of TEsorter?

[Juke34]

Sure, the container has to be build after the modification. And as the EDTA container is made only when you release a new version of EDTA, if you did not release any new EDTA version since the hmmpress modification the EDTA container cannot embed the last version of the hmmpress dependency.