I'm trying to run Rosetta to do a fold and dock of the C-terminal of SARS-CoV-2 spike protein.
When I run make_fragments.pl for the first time, it downloads a bunch of nr*.gz files, and then fails with the following error:
[fastacmd] ERROR: ERROR: Cannot initialize readdb for nr database
This is a new Rosetta install of the latest version.
Digging into this a bit, make_fragments.pl runs install_dependencies.pl which runs fastacmd -D 1, like this:
my $cmd = "$Bin/blast/bin/fastacmd -D 1 > $datdir/nr";
In my case the data is in /mnt/data/rosetta/tools/fragment_tools/databases so I tried changing to that directory and running /mnt/data/rosetta/tools/fragment_tools/blast/bin/fastacmd -D 1 directly; same error.
The databases directory is rather large, 217GB. It contains nr.00.phd nr.00.phi nr.00.phr nr.00.pin nr.00.pog nr.00.ppd nr.00.ppi nr.00.psq nr.00.tar.gz.md5 and so on, for nr.00 through nr.38; also nr.pal nr.pdb nr.pos nr.pot nr.ptf nr.pto. It doesn't seem like there were any issues with downloading and unpacking the nr database.
I'm not sure why fastacmd produces an error or how to fix it. What is it looking for? Why is the thing it's looking for not there?
Environment: Rosetta 2020.08.61146 (rosetta_bin_linux_3.12_bundle.tgz), Ubuntu 18.04 on AWS instance
Thanks! How recent of a change is this? My Rosetta seems to be from March 9, 2020.
It looks like the nr download is handled by ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.10.0/ncbi-blast-2.10.0+-src.tar.gz in c++/src/app/blast/update_blastdb.pl (it was expecting ncbi-blast-2.9 but couldn't find it so I had to bump the version) and fastacmd is from ftp://ftp.ncbi.nih.gov/blast/executables/legacy.NOTSUPPORTED/2.2.17/blast-2.2.17-x64-linux.tar.gz
This change was done on Feb 4th, 2020: New layout for NCBI BLAST FTP database site starting February 4, 2020
You may need to use blastdbcmd instead (sounds like that is retrieving fasta sequences).
Hi, could you please describe how did you sapped to v4? I'm having the same problem.
Thanks,
Hi, could you please describe how did you sapped to v4? I'm having the same problem.
Hello Diego,
I did something like this: In a clean rosetta install, edit main/tools/fragment_tools/install_dependencies.pl to comment out these four lines
# system("perl $datdir/update_blastdb.pl nr");
# system("perl $datdir/update_blastdb.pl nr");
# system("perl $datdir/update_blastdb.pl nr");
# (system("perl $datdir/update_blastdb.pl nr") == 0) or do { &clean_nr_tgz; };
Run these commands to download v4 (if you don't have wget, install it first)
mkdir main/tools/fragment_tools/databases/
cd main/tools/fragment_tools/databases/
wget 'ftp://ftp.ncbi.nlm.nih.gov/blast/db/v4/nr*gz'
and NOW you can run install_dependencies.pl as usual
Thank you very much.
I'm going to try this.
Best regards
Another note for someone who ends up here. for fastacmd to be able to find nr, the variable BLASTDB must be set.
try two things to figure out whether this is the issue:
echo $BLASTDB should return the path to database.
if it is not set, set it manually:
export BLASTDB=$(pwd)/databases
then try getting database info from fasatacmd, before going to run the long nr generation:
blast/bin/fastacmd -d nr -I
If you get the info here, then change nr version, but if your BLASTDB is not set correctly and you don't even get info, make sure to set it.
Log in to answer this question.
