htseq-count produces no features amid alignment

Question

[bojingjia]

I recently aligned some sequencing data using STAR against an mm10 prebuilt genome. Afterwards, I sorted and indexed using samtools, and proceeded to generate read counts using htseq-count (and appropriately, an mm10 Ensembl gtf file). But all of my read counts are peculiarly all 0's, classified as no features.

A number of other users have reported the same problem here on BioStars, but their concerns weren't resolved. I have to wonder if my reads failed to align, but a quick look at the bam files in IGV shows many aligned reads. Am I using a faulty mm10 annotation file? Would anyone have suggestions/comments?

[Amitm]

hi,

A couple of checks -

1. Default sorting order expected by HT-Seq in the BAM is name. Most aligners return coord. sorted BAM
2. You aren't using a GTF file downloaded from UCSC. Last time I checked it had conflict in gene_id with transcript_id values. Read more in the FAQs at the end of this page.
3. Chr name style is same in your BAM and GTF

Answer 1

[ablanchetcohen]

Do the sequence names in the BAM file and the GTF file match?

UCSC and Ensembl use different chromosome nomenclatures.

I will never understand why we can sequence the human genome, and put men on the moon, but not agree whether chromosome 1 should be referred to as chr1 or 1.