strange error

[Georgi Marinov]

Hi,

I am trying to run GRIT and I get this message:

  File "/usr/bin/run_grit.py", line 514, in <module>

    main()

  File "/usr/bin/run_grit.py", line 486, in main

    elements = discover_elements(sample_data, args)

  File "/usr/bin/run_grit.py", line 440, in discover_elements

    sample_data.get_reads(sample_type)

  File "/usr/bin/run_grit.py", line 247, in get_reads

    rnaseq_reads = self._load_rnaseq_reads(sample_type, rep_id)

  File "/usr/bin/run_grit.py", line 188, in _load_rnaseq_reads

    assert data.paired == 'true', "RNASeq reads must be paired"

AssertionError: RNASeq reads must be paired

This happens on both TopHat and STAR-generated BAM files.

Does this mean that there is a requirement for all reads to be paired and that if you have singletons, the program exits or there is some other problem? 

Thanks,

Georgi

[Nathan Boley]

> This happens on both TopHat and STAR-generated BAM files.
>
> Does this mean that there is a requirement for all reads to be paired and
> that if you have singletons, the program exits or there is some other
> problem?

No, but it does ignore RNAseq reads without proper pairs. It looks
like it's choking during the argument parsing, rather than when
processing the reads.

Are you using a control file or command line options? Can I see the
control file/exact command?

Best, Nathan

[Nathan Boley]

[cc-ing list]

> python /usr/bin/run_grit.py --threads 8 --rnaseq-read-type backward
> --rnaseq-reads wgEncodeCshlLongRnaSeqMcf7CellPapFastqRd1Rep1.STAR_2.3.0e.bam
> --ofprefix
> GRIT-1.1.0-wgEncodeCshlLongRnaSeqMcf7CellPapFastqRd1Rep1.STAR_2.3.0e &
>

It looks like I introduced another bug when I changed the argument
parsing. I'll upload a fix tonight.

Also, you'll need to either provide poly(A) and CAGE or RAMPAGE reads,
or reference poly(A)'s and promoters.

> What exactly is the CONTROL file supposed to contain and do? I could not
> find any example so I did not specify any

For now, you can use a control file of the form:

# comment line
# *'s indicate merged
#sample_type rep_id assay paired
stranded read_type filename

AdMatedF_Ecl_20days_Heads rep1 rnaseq true true
auto AdMatedF_Ecl_20days_Heads.biorep1.rnaseq.chr4.bam
AdMatedF_Ecl_20days_Heads rep1 cage false true
auto AdMatedF_Ecl_20days_Heads.biorep1.cage.chr4.bam
AdMatedF_Ecl_20days_Heads rep1 polya false true
auto AdMatedF_Ecl_20days_Heads.biorep2.passeq.chr4.bam

I've put an example data set with a control file and README at:
http://grit-bio.org/GRIT_example.tar.gz

Note that to use read type 'auto' you'll need to provide a reference
annotation (even GENCODE chr20 will work - GRIT just needs genes so
that it can determine the gene strand).

> P.S. Am I correct in my guess the read-type should be "backward" for dUTP?

Yes.

Thanks for your interest - sorry about the bug.

Best, Nathan