combining htseq-count files into one

[mbk0asis]

Hi.
I have 12 htseq-count result files.
When I tried to combine them with Excel, I found that there were some missing rows in one file that were present in another.
For example,
<file1>
a1 3
a2 4
a3 5
a4 3

<file2>
a1 4
a3 6
a4 5


Hot can I combine without loss of data like following?
<file 1+2>
a1 3 4
a2 4 N/A
a3 5 6
a4 3 6


Thank you!

[Simon Anders]

1. If there are missing rows, then you have a prime example for a data manipulation that is trivial in any statistics or scripting language but really complicated in Excel. Time to learn some scripting.

2. There shouldn't be any missing rows, because htseq-count outputs a count for each gene that is mentioned with an exon line in the GTF file, no matter whether reads map to it or not. Something went wrong. Maybe you used different annotation files?

[mbk0asis]

Thanks for the reply.
I'm actually learning python now, but it is difficult learn it since no one around me is familiar with scripting.
Anyway, I ran htseq-count with SAM and GTF files for each sample.
Following is the command I used for it.

htseq-count -m intersection-strict -s no -i gene_naem I1.sam /home/bk/Desktop/bosTaurus_2013/GTF/I1_transcripts.gtf > I1_counts.txt

I have data from 12 different samples, so I changed the file names according to the sample names.

[chadn737]

Are these de novo assemblies or did you make a gtf file for each sample using Cufflinks? Because it looks like you are using a gtf file specific to the sample.

In that case, each gtf will have slightly different lists of transcripts. In order to have htseq-count give the same output for each sample, you need to use the same gtf file for all of them.

[mbk0asis]

I didn't make GTF files. They are from a company we paid for sequencing.
I know the company used Cufflinks for expression estimation.

They didn't really explain me about those files. Besides I don't have much knowledge in the bioinfo. area, so everything is confusing for me.

So what should I do to get the same GTF file

Sorry for my disorganized question. English is not my primary language, so please understand...

[dpryan]

Have a look at the cuffmerge command that comes with cufflinks.

[chadn737]

Not a problem. Use cuffmerge to merge all the GTF files into a into a single GTF and then use htseq-count using the merged GTF file.

http://cufflinks.cbcb.umd.edu/manual.html#cuffmerge