[shi]

I don't understand how you solved the problem by changing the sam file? The problem is with your GTF annotation file, not your sam file.

[chibouki]

Finally, we solve the problem by cutting the last column of sam file.

But we are disappointing, it seems that -d and -D options are only for pair-end?

[shi]

Hello,
I think I may have a problem with my gtf file, which look like this:
##gff-version 3
##source-version geneious 6.1.6
seq_ref Geneious gene 1 109 . . . Name=exon1;created by=User

I received this error message:
Warning: failed to find the gene identifier attribute in the 9th column of the provided GTF file.
The specified gene identifier attribute is 'exon'
The attributes included in your GTF annotation are 'Name=exon1;created by=User;modified by=User'

I hope you could help me

Your GTF file has an incorrect format. Have a look at this page for GTF format:

http://mblab.wustl.edu/GTF2.html

[shi]

Hi Wei,
I may have found a bug. It seems that the last line of my BAM file header is being parsed as the fist read, since it is quite long this causes the program to exit (line 697 in core.c). When I remove the last @PG line from the header and reheader the bam file your code runs to completion. Please see the attached header.
cheers,
Aaron

Dear Aaron,

Thanks for reporting this. You are correct that there was a bug in featureCounts in dealing with long header lines. We have fixed this and released a patched version of Subread package (1.4.3-p1).

Best wishes,
Wei

[chibouki]

Hello,
I think I may have a problem with my gtf file, which look like this:
##gff-version 3
##source-version geneious 6.1.6
seq_ref Geneious gene 1 109 . . . Name=exon1;created by=User

I received this error message:
Warning: failed to find the gene identifier attribute in the 9th column of the provided GTF file.
The specified gene identifier attribute is 'exon'
The attributes included in your GTF annotation are 'Name=exon1;created by=User;modified by=User'

I hope you could help me

[AntonioMaceo]

SAM/BAM parse error

Hi Wei,
I may have found a bug. It seems that the last line of my BAM file header is being parsed as the fist read, since it is quite long this causes the program to exit (line 697 in core.c). When I remove the last @PG line from the header and reheader the bam file your code runs to completion. Please see the attached header.
cheers,
Aaron

[shi]

We have just released Subread 1.4.2. The featureCounts program included in this release outputs an assignment summary file (*.summary) along with the read count file.

BTW, our featureCounts paper was just published on Bioinformatics. Here is the link to it:

http://bioinformatics.oxfordjournals...6F&keytype=ref

Wei

[bw.]

Thank you Wei. I switched to using featureCounts.
Its especially great that it now takes multiple .bam files.

Whatever statistics can be provided in future versions will be helpful.

[shi]

What Bruce suggested is the current way of getting assignment stats from using featureCounts. But apparently it is not user-friendly. It is on our to-do list to output assignment stats (with no need to turn on -R option) when read summarization is done, but it will take a week or two to implement it. We will make a new release in a day or so, but it is not going to be included in that release.

Best,
Wei

[bruce01]

Ben, I had the same issue, so made a command to get this info. It requires you to make the 'reads' output using -R flag.

cut -f 2 <featco.counts.reads> | sort | uniq -c > <featco.counts.diags>

Output looks like:

154266 ACCEPTED_2VOTE_GENE
23169444 ACCEPTED_GENE
40066 MULTI_MAPPING
4470627 NOTFOUND_GENE
100013 OVERLAPPED_GENES
2850 PAIR_DISTANCE

Hope that helps, Bruce.

[bw.]

Hi,
I would like to use featureCounts, but miss the stats provided by htseq-count (copied below) as these let me make sure I got the 'strand' setting right and other things.
Any chance you could add similar output to featureCounts (either as a separate 'stats.txt' file or as part of the main table)?

no_feature 20123817
ambiguous 9026940
too_low_aQual 0
not_aligned 0
alignment_not_unique 3034042

Thanks
-Ben

[shi]

Hi, quick question. I was wondering if there was a way to get featureCounts to work on a Windows 7 OS. Going through R and Bioconductor would be perfect, but it looks like Rsubreads does not have a Windows version? Is there any other way?

You are correct that Rsubread does not have a Windows version. It is pretty hard to develop a Windows version for this package due to most of the code was written in C. I think we might eventually come up with a Windows version, but it will take a fair bit of time. If you have access to a unix machine, you can fairly easily use featureCounts via the Bioconductor package Rsubread.

Wei

[shi]

Dear Wei,

I tried that command line, but it still drops a Segmentation fault. I also tried aligning my reads using Subreads (which succeeded), but when I ran featureCounts on the resulting SAM file I also got a Segmentation fault. The output is the same as I attached to a previous post.

Any more ideas?

Hi,

Thank you for trying these options. We found featureCounts always works nicely with Subread. So the segment fault is likely to be due to some unexpected data in the annotation. We have also received some other bug reports similar to this recently. The 1.3.x version of featureCounts allows up to 60 features overlapping with each other in the annotation. If the number of such features exceeded this limit, we found the program crashed. Although this is rare but it may happen and we suspect this might be the reason causing the seg fault seen in your data.

We have removed this limit in the latest version 1.4.0 and hopefully this will solve the problem.

Also, if reads in your BAM file were sorted by chromosomal locations, you should include '-S' option in your command. Not doing so will not crash the program, but will result in incorrect read counts.

Let me know if the problem persists.

Wei

[adaigle]

Hi, quick question. I was wondering if there was a way to get featureCounts to work on a Windows 7 OS. Going through R and Bioconductor would be perfect, but it looks like Rsubreads does not have a Windows version? Is there any other way?

[choseqid]

Dear Wei,

I tried that command line, but it still drops a Segmentation fault. I also tried aligning my reads using Subreads (which succeeded), but when I ran featureCounts on the resulting SAM file I also got a Segmentation fault. The output is the same as I attached to a previous post.

Any more ideas?