Open lxsteiner opened 4 years ago
Attached is what directories should look like after a successful run:
Those error messages are bad, meaning none/little of the tRNA/tmRNA-finding is working; Islander would then be unable to produce any results. It might help the troubleshooting if you post a similar file of directory structure to what I posted above. (Even deeper, into the subdirectories of the first "trna" would be better.)
Thanks for providing an output example.
I've been still fighting with getting all dependencies to work. Here are the latest issues I hit with islander.pl
Possible attempt to separate words with commas at /home/leon/tools/TIGER/bin/rfind.pl line 22.
which does seem to be an issue in the script because line 22 in rfind.pl
looks like this:
20 #my @blastCmd = (qw/blastn -soft_masking false -lcase_masking -db/, $dbfiles, qw/-outfmt 6/); # Run3 wants an array of its command words!
21 my @blastCmd = (qw/blastn -dust no -soft_masking false -lcase_masking -db/, $dbfiles, qw/-outfmt 6/); # Run3 wants an array of its command words!
22 my @mergeCmd = qw/bedtools merge -s -c 4,6 -o collapse -i stdin/;
23 #warn scalar(keys %inseqs), " inseqs, first of size ", length($inseqs{(keys %inseqs)[0]}), "; db=$dbfiles\n";
I think there are ()
missing for the array on line 22, no? Haven't used Perl in a long time.
FATAL: Unable to find /usr/local/bin/cmsearch executable
something is specifically calling a global installation of cmsearch, I thought that it was maybe hardcoded somewhere accidentally but the only script in TIGER where I can find it is introns.pl
in the following lines:
376 #my $cmd = "cmsearch --cpu 0 --tblout gpi.tbl $lib/cm/Intron_gpI.cm $file &> /dev/null";
377 my $cmd = "cmsearch --cpu 0 --tblout gpi.tbl $lib/cm/gpi_bact_trna.cm $file &> /dev/null";
378 RunCommand($cmd, 'gpi.tbl');
and I don't see why it would explicitly call /usr/local/bin/cmsearch
because I have binaries exported as for everything else as well and running fine:
$ which cmsearch
/home/leon/tools/infernal-1.1.3-linux-intel-gcc/binaries/cmsearch
is introns.pl
even called from islander.pl
? Not sure if it's then some other piece of software doing it...
I'm also getting some errors with Prokka (1.13) which I don't think are related to TIGER:
Argument "1.7.7" isn't numeric in numeric lt (<) at /home/beate/anaconda/bin/prokka line 253.
Use of uninitialized value in concatenation (.) or string at /home/beate/anaconda/bin/prokka line 201.
Use of uninitialized value in numeric lt (<) at /home/beate/anaconda/bin/prokka line 202.
[12:05:35] Prokka needs signalp 3.0 or higher. Please upgrade and try again.
it's probably some BioPerl conundrum... also it asks for singalp, but signalp is there:
$ signalp -version
SignalP version 5.0b Linux x86_64
I think I'll try upgrading Prokka to the newest version and hopefully some of these errors will be solved then. But please do comment on the 1st and 2nd error I mentioned here.
Thanks!
Interesting development of events. After upgrading Prokka, the Prokka issues are gone and I'm getting a little bit more output from islander.pl
but nowhere near the entire directory structure from your output example and some files still empty.
The issue with cmsearch still persists even though I have cmsearch in a conda environment and locally outside of it.
FATAL: Unable to find /usr/local/bin/cmsearch executable
which is funny because there is even a global cmsearch installation although an older version:
$ /usr/bin/cmsearch -h
# cmsearch :: search CM(s) against a sequence database
# INFERNAL 1.1rc4 (June 2013)
but the correct versions are all exported to the path before the global one:
$ which cmsearch
/home/leon/tools/infernal/binaries/cmsearch
or in conda:
$ which cmsearch
/home/leon/miniconda3/envs/PROKKATIGER/bin/cmsearch
also this time the islander.log was 36 MB big because it includes an actual output from cmsearch and hmmsearch.
Here is the entire output if you want to check it out: islander.zip
I'm not sure what to try and change anymore :/
I'm getting several errors when running the first step with
islander.pl
on thetestdata
and I'm not sure if it's normal because there's also a lot of "completed successfully".Notable ones:
please see full islander.log in attachment. It finishes with:
so not sure if this is ok or not.
I installed all dependencies with conda to match your exact versions:
The only difference is tRNAscan-SE which can only be 2.0-1 or 2.0.3, because 2.0.2 is not available.
Thanks, Leon