[maker-devel] maker-2.23 with tophat / cufflinks input

[Carson Holt]

The error is caused by a perl module that is loaded at runtime to access
the system's Process Table.  This is an error that sometimes happens in
cluster environments like SGE.  It's not related to the GFF3.  It's more
related to the specific system running MAKER at the time and can be
semi-random.  MAKER has auto-retry on contigs, so it would have done it's
best to recover from any failure.  If the master_index_datastore.log file
shows all input contigs as being of status FINISHED, you are fine.

If you are on a cluster and you continually get the error you may need to
install the Proc::ProcessTable locally on that specific node and not just
globally for the cluster.  Errors with fold_constants happen when you
compile on one system and then run on another.  The issue can sometimes be
related to floating points being slightly different between the two
systems.  So clusters that have non-identical architectures within the
same cluster can have these types of errors more often.

Thanks,
Carson



On 12-02-23 12:16 PM, "Olaf Mueller" <olaf.mueller at duke.edu> wrote:

>Hi,
>
>I've done a preliminary run on an alga genome with maker 2.23beta, using
>only mRNA-seq tophat/cufflinks input and no additional evidence source.
>
>As prerequisite a cufflinks transcripts.gtf (merged from different read
>sample mappings) has been converted with maker cufflinks2gff3
>(strandless transcripts/exons removed). I provided the resulting gff3
>via pred_gff and model_gff to maker, and also didn't call est2genome or
>other predictors.
>
>During maker run for some scaffolds, I receive this error message for
>some chunks, while gathering GFF3 input into hits:
>
>Gathering GFF3 input into hits - chunk:0
>panic: fold_constants JMPENV_PUSH returned 2 at
>/home/muol/Documents/Software/maker/lib/File/..//Proc/Signal.pm line 89
>Gathering GFF3 input into hits - chunk:2
>...
>
>The index log nevertheless reports these scaffolds as finished and
>annotations are generated. Do I have to be concerned about this message
>because it is related to my cufflinks input?
>
>Thanks
>Olaf
>
>
>_______________________________________________
>maker-devel mailing list
>maker-devel at box290.bluehost.com
>http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org