Cuffmerge error - please help!

[yehudithasin]

Hi all
I need your help please (and yes, I read the previous cuffmerge error posts, and did not get a solution thet works). I need to run cuffmerge ~250 files, but to test it first I am trying to run it just on two, and get the following error:
********************
[Thu Jun 19 14:44:59 2014] Beginning transcriptome assembly merge
-------------------------------------------

[Thu Jun 19 14:44:59 2014] Preparing output location test211/
Traceback (most recent call last):
File "/u/local/apps/cufflinks/2.1.1/cuffmerge", line 580, in <module>
sys.exit(main())
File "/u/local/apps/cufflinks/2.1.1/cuffmerge", line 546, in main
chrom_info = get_gtf_chrom_info(gtf, chrom_info)
File "/u/local/apps/cufflinks/2.1.1/cuffmerge", line 476, in get_gtf_chrom_info
left = int(cols[3])
ValueError: invalid literal for int() with base 10: 'transcript'
*************

I tried to use several different reference gtf files (all downloaded from UCSC) or run without reference (just merge the transcript files) and get the same error.
The cuffmerge command is:
cuffmerge -o test -g PATH2/GENECODEM2_mm10_comprehensive_processed.gtf -s PATH2/Chromosomes -p 8 assembly_list.txt

The assembly file is:
/u/scratch/y/yhasin/StrainBAMsorted/Sample_SWR_J_cuff/transcripts.gtf
/u/scratch/y/yhasin/StrainBAMsorted/Sample_129X1_SvJ_cuff/transcripts.gtf

Any suggestions?

Thank you
Yehudit

[yehudithasin]

Ok, so I found the problem (thanks to adiallo's advice), and its really weird:
Cufflinks apparently made a slip in some of the files, and one row in the transcript.gtf file of *some* samples looks like this:
"chr4 Cufflichr4 Cufflinks transcript 137468769 137570630 1 + . gene_id "CUFF.10499"; transcript_id "ENSMUST00000030547.8"; FPKM "0.0000000000"; frac "0.000000"; conf_lo "0.000000"; conf_hi "0.000000"; cov "0.000000"; full_read_support "no";"
i.e. there is an extra field in one line....

while all the rest look normal (i.e. like:
"chr4 Cufflinks exon 140765571 140765637 1000 - . gene_id "ENSMUSG00000025330.6"; transcript_id "ENSMUST00000026381.6"; exon_number "14"; FPKM "0.0259126396"; frac "1.000000"; conf_lo "0.025913"; conf_hi "0.025913"; cov "0.092702";"

In case you are running into this - I found the line number by getting all the field 4 values (awk <transcript.gtf '{print $4}' >temp) and then "grep -n 'transcript' temp" to find the line number....

This is really weird, and the first time I see something like this for cufflinks (I used version 2.2.1 to generate the files, and STAR to align)