This is a test version of Biostars. For the public version, visit https://www.biostars.org.
How to set hhm_db and a3m_db for HHPRED?

I've been running HHPRED, and following error message appears:

  • Could find neither hhm_db nor a3m_db!

How can I configure these variables on HHPRED? Github page [1] says nothing about it!!

[1] https://github.com/soedinglab/hh-suite/tree/master/scripts/hhpred

For those who have ResearchGate account, this question is also available over there: https://www.researchgate.net/post/How_to_set_hhm_db_and_a3m_db_for_HHPRED

software error structure hh-suite hhpred
• 4,573 views • link
updated by Joe Joe • written by inaciogmedeiros • inaciogmedeiros

Have you downloaded the databases from their site and set up the environment variables you need?

Installing the suite is not the most straightforward process it has to be said. It might have gotten better since it moved to github but im not sure.

Take a look at the full PDF manual, it has considerably more detail. Im not at my computer right now to tell you the exact steps you need, but if no one beats me to it i'll do so later.

• 0 views • link
written by Joe Joe

1 answer

Their github does say something about it, you need to take a look at their README.

I think you can ignore my previous comment about environment variables etc. I'm still running the older version of HHsuite (before it moved to Github, and before version 3 came out. It looks (to my eye at least), like they've considerably streamlined the process of setting up databases, though the same general principle applies:

  1. Download the database you want from their webserver:

    wget robots=off -r --no-parent -nH -nd -np -R *.html,*.txt -A .tgz http://wwwuser.gwdg.de/~compbiol/data/hhsuite/databases/hhsuite_dbs/

Will download ALL the databases (be warned, they're ENORMOUS files).

  1. Extract them:

    tar -xvzf /path/to/database.tgz -C /path/to/folder

The -C ... is optional if you downloaded them where you wanted them in the first place.

  1. The rest of their instructions should be easy enough to follow, you are simple adding the file paths to HHSUITE's config files so that it knows where to look for them:

    - enter the path to the database into the config.pm file in the "lib" folder (entry: _pdbdir)

- enter the path to the uniprot database into the config.pm file in the "lib" folder (entry: _uniprot20)

- create a folder called "db" in your database folder, open a terminal, change into this folder and call "cat ../*.hhm > pdb.hhm"

Hopefully that's everything. I've not yet gotten around to testing out their V3.0 so this answer might not be fully complete.

• 0 views • link
written by Joe Joe

That's the question: I've followed every step guidelined in README: I've downloaded the databases, I've set the environment variables, and so on. "hhm_db" and "a3m_db" are variables used in hh-pred source code, certainly related to databases that README guides. How to set them is my doubt.

Another question: It is said "and extract them into your hhpred database folder". This database folder can be any folder I want, so I just need to say which is in step "enter the path to the database into the config.pm file in the "lib" folder (entry: _pdbdir)"?

• 0 views • link
written by inaciogmedeiros • inaciogmedeiros

I'm not sure what is up with their Github exactly. I've just cloned and installed the version at that link and it still claims to be V2.0.15.

There is a weird quirk depending on which process you're running that caught me off guard when I first started using it, perhaps you have the same issue: if you're using hhsearch for example, you need to specify the exact file for the database:

$ hhsearch -d /path/to/pdb70_hhm.ffdata

However with some programs, you need to only specify the 'basename' of the database, since it depends on several files with the same root and different extensions.

What exact command are you trying to run?

• 0 views • link
written by Joe Joe

I'm trying to run hhpred. Looking at its "help" when executing it, and on hh-suite userguide, there's nothing talking about set a "database flag" to hhpred.

• 0 views • link
written by inaciogmedeiros • inaciogmedeiros

I'm not sure what you mean by hhpred in this instance, as the commandline programs are hhblits and hhsearch AFAIK. There are other helper commands like hhmake, but blits and search are the actual search programs. HHpred is their name for their webserver implementation I'm fairly sure.

Can you copy an exact command so that I can see what you're trying to do?

I routinely use the software like so for instance:

hhsearch -dbstrlen 50 -B 1 -b 1 -p 60 -Z 1 -E 1E-03 -nocons -nopred -nodssp -cpu 25 -i "${file}" -d "${db}"

To just get the top most hits (I don't care much about the alignments etc)

In this case, $file is used as I loop the program over several hundred proteins, and $db evaluates to the result of a find command, so that I don't have to specify the database directly (every time I download a new one the name changes).

• 0 views • link
written by Joe Joe

Can you copy an exact command so that I can see what you're trying to do?

Sure yeah, here it is (where "." refers to hh-suite directory [I've downloaded and compiled from GitHub repository]):

./scripts/hhpred/hhpred.pl -i seq.fa -o structure.pdb > output
• 0 views • link
written by inaciogmedeiros • inaciogmedeiros

did you eventually find how to run hhpred?

• 0 views • link
written by jfloresu • jfloresu

Sorry I totally missed this reply/comment.

Try running the binary that should be in a /bin directory instead of the Perl script if memory serves (unfortunately I can’t check my set up right now).

• 0 views • link
written by Joe Joe

Hi folks. Since then, I've redirected my activities to other branches, although I'm expecting to return to HHPred soon.

I'm gonna check if there's a bin directory, and hope that HHPred has been released new stable versions available for download (last time it was a beta version).

• 0 views • link
written by inaciogmedeiros • inaciogmedeiros

Log in to answer this question.

More posts like this