HTSeq count to expression level

stormin · 06-15-2015, 05:26 PM

Hi,

I understand that the raw HTSeq output is number of reads mapped to each gene (correct me if I'm wrong). Is there an easy way of converting counts to gene expression? The method should take into account gene length as well as total seq depth to be comparable between samples.

Also, I use DESeq2 to do differential expression using HTSeq raw output, is that a valid method? Does DESeq2 automatically take into account gene length?

Thank you!

dpryan · 06-16-2015, 12:14 AM

If you really wanted to, you could convert the counts to RPKM or FPKM, though you shouldn't do statistics on those.

Alternatives to DESeq2 include edgeR and limma/voom, both of which will accept the same counts. DESeq2 doesn't account for gene length because it's constant between group. If you have a gene length bias between samples that's confounding things then look at the CQN package.

stormin · 06-16-2015, 10:32 AM

Quote:

Originally Posted by dpryan

If you really wanted to, you could convert the counts to RPKM or FPKM, though you shouldn't do statistics on those.

Thanks for the concise answer, are there any scripts that can do this conversion?

dpryan · 06-16-2015, 10:34 AM

I think edgeR has a function to do that if you supply the gene lengths. That's such a simple thing to compute that I suspect most people just write a script as needed.

stormin · 06-16-2015, 11:01 AM

Sounds good, thanks!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
htseq-count	moon_ligth	Bioinformatics	2	11-15-2014 05:26 PM
htseq-count low count problem	gandalf886	Bioinformatics	3	08-23-2014 08:05 AM
HTSeq exon counts to gene level	Minty	RNA Sequencing	10	06-18-2014 05:48 AM
HTseq-count for antisense expression	BacanBan	RNA Sequencing	0	10-25-2012 06:47 AM
multiBamCov or htseq-count to count read per feature ?	NicoBxl	Bioinformatics	1	07-03-2012 03:05 AM

06-15-2015, 05:26 PM	#1
stormin Member Location: US Join Date: Aug 2014 Posts: 23	HTSeq count to expression level Hi, I understand that the raw HTSeq output is number of reads mapped to each gene (correct me if I'm wrong). Is there an easy way of converting counts to gene expression? The method should take into account gene length as well as total seq depth to be comparable between samples. Also, I use DESeq2 to do differential expression using HTSeq raw output, is that a valid method? Does DESeq2 automatically take into account gene length? Thank you!

06-16-2015, 12:14 AM	#2
dpryan Devon Ryan Location: Freiburg, Germany Join Date: Jul 2011 Posts: 3,480	If you really wanted to, you could convert the counts to RPKM or FPKM, though you shouldn't do statistics on those. Alternatives to DESeq2 include edgeR and limma/voom, both of which will accept the same counts. DESeq2 doesn't account for gene length because it's constant between group. If you have a gene length bias between samples that's confounding things then look at the CQN package.

06-16-2015, 10:34 AM	#4
dpryan Devon Ryan Location: Freiburg, Germany Join Date: Jul 2011 Posts: 3,480	I think edgeR has a function to do that if you supply the gene lengths. That's such a simple thing to compute that I suspect most people just write a script as needed.

06-16-2015, 11:01 AM	#5
stormin Member Location: US Join Date: Aug 2014 Posts: 23	Sounds good, thanks!