CEGMA error

[flobpf]

Hi,

I've been trying to run CEGMA on my 280MB genome. It does run fine on all other genomes I've got, but it refuses to run on my genome. It keeps giving the following error:

RUNNING: local_map -n local -f -h /opt/cus/CEGMA/2.3--GCC-4.1.2/data/hmm_profile
s -i KOG -v  genome.chunks.fa 2>sample.cegma.errors
FATAL ERROR! 6400: "No such file or directory"

In sample.cegma.errors
Warning Error
        In GeneLoop6 matrix special read off - out of bounds on matrix [i,j
is 193,-1 state 0 in standard matrix]
Warning Error
        Problem - hit bad read off system, exiting now
Can't use an undefined value as an ARRAY reference at /opt/cus/CEGMA/2.3--GCC-4.
1.2/bin/parsewise line 217.
Can't run parsewise

It is weird that this problem occurs with only my genome. I've tried removing all the non-canonical characters and changing all masked sequences to upper case. However, it does not work.

I wonder if anyone has encountered anything like this? If yes, how did you solve it, if at all?

Thanks

[pengchy]

https://gist.github.com/1153173

[flobpf]

Actually, CEGMA's most recent version did just fine. My problem was resolved thanks to the deft bug-fixing by the CEGMA group.

[xushengxun]

Problem when using CEGMA

Actually, CEGMA's most recent version did just fine. My problem was resolved thanks to the deft bug-fixing by the CEGMA group.

Hi flobpf,
I encountered the same problem as you when using CEGMA. The error is as follows:

Warning Error
Major problem (!) - in GeneLoop6 matrix to special read off, position
0,0 state 0 no source found!
Warning Error
In GeneLoop6 read off ending at 0 ... got a bad matrix to special
read off... returning partial alignment
Can't use an undefined value as an ARRAY reference at /state/partition1/home/xuchao/software/cegma_v2.3.190711/bin/parsewise line 217.
Can't run parsewise

You seem to have solved this problem! Could you tell me how?

[AdrianP]

Actually, CEGMA's most recent version did just fine. My problem was resolved thanks to the deft bug-fixing by the CEGMA group.

Bold is solution.

[aliceb]

Hi all,

This is an old thread, but I'm getting a similar error with CEGMA v2.5. It's not happening all the time - I ran the test data just fine, the test proteins against my data ok, and my data against the Anopholes data ok. But my data against the Core proteins (instead of the 248 Anopholes ones) crashes.

Is anyone else having this problem?

Here are the last few lines:
********************************************************************************
** ACCURATE LOCAL MAPPING **
********************************************************************************

RUNNING: local_map -n local_self -g local.genewise.gff -d geneid_params/self.param -h /media/data/programs/CEGMA_v2.5/data/hmm_profiles -i KOG genome.chunks.fa 2>../annot_core/annot_core.cegma.errors
NOTE: Will use specifed local.genewise.gff file instead of running genewise
FATAL ERROR when running local map 6400: "No such file or directory"

Ending CEGMA

#############
Thanks,
Alice

[kbradnam]

Does the output before this stage imply that you have no core genes? I.e. sometimes when you run CEGMA, the initial BLAST step might only find a small number of core genes. Each successive filtering step of CEGMA typically throws out more candidate core genes and you can end up with none at all. Some of the code might not handle this gracefully.

[aliceb]

Yes, I did think of that. But I have now been able to get 99% hits on a dataset that crashes against the Anopholes core genes. It seems like there should be some hits there (I am working with a wasp)....

I have continued to play my two (fasta, nucleotide) datasets and have managed to successfully run one against the core genes and the other against the Anopholes core genes. This is good enough for now, but a bit frustrating as I wanted to compare them....

If anyone comes across this and has a solution, it would still be useful for me in the future (and perhaps other people hitting this snag)

[timothylaurent]

crashing on 2.5

I have Cegma 2.5 freshly installed and I am getting similar errors:


Would love to be able to run the software....


********************************************************************************
** MAKING INITIAL GENE PREDICTIONS FOR CORE GENES (GENEWISE + GENEID) **
********************************************************************************

RUNNING: local_map -n local -f -h /build/CEGMA_v2.5/data/hmm_profiles -i KOG genome.chunks.fa 2>/out/S376-cegma.cegma.errors
FATAL ERROR when running local map 6400: "No such file or directory"

Ending CEGMA




In my .cegma.errors file it says this:

Warning Error
Unable to open gene.stat as gene stats file
Warning Error
Could not read gene statistics in gene.stat
Warning Error
Could not open BLOSUM62.bla as a filename for read Blast matrix
Warning Error
Could not open file codon.table as codon table file
Warning Error
Could not read codon table file in codon.table
Fatal Error
Could not build objects!
Can't run genewise -splice_gtag -quiet -gff -pretty -alb -hmmer /build/CEGMA_v2.5/data/hmm_profiles/KOG0002.hmm /tmp/genomic18349.fa > /tmp/genewise18349

[timothylaurent]

memo to self -- RTFFAQ


The FAQ talks about this being due to a

genewise ocnfig being installed incorrectly.

I installed wise on an ubuntu container with apt-get

this should fix the problem in my running container:

export WISECONFIGDIR=/usr/share/wise

and this should fix it in my Dockerfile:

ENV WISECONFIGDIR /usr/share/wise




Maybe this is the root of some of the other problems.

[kbradnam]

As time marches along, there will only ever be more problems like this because CEGMA has been locked in development using some older versions of tools. As people install newer versions of things like HMMER and genewise (well, maybe that's not a tool that will be updated much either), things are likely to break.

If we ever get new funding, we have many plans for a completely new version of CEGMA, but without money we can't spend any more time on it unfortunately.

Regards,

Keith

[timothylaurent]

So the environment variable didn't help with the problem . ( although the command would run if I put it on the command line ).


Luckily the docker container approach comes configured. Don't waste time trying to get this installed on your system... just use the docker container -- then it is just a 1 liner

[colindaven]

The VM works well too, haven't tried the docker container yet.

One word on whole genomes - every contig or scaffold set has been fine so far, but running it on whole genomes with long chromosomes has failed repeatedly.

Should long chromosomes be split into arbitrary scaffolds ?