RNA SEQ Bias toward short transcripts (RPKM)

vruotti · 01-06-2010, 08:19 AM

Hi there.
Just wanted to post this to see if anyone else see this interesting pattern. When taking the average expression (RPKMs) of short genes, say 0.5kb to 2.5kb, we see that the average is much higher when comparing to the average of longer genes, say 10.5-50.5kb. We have done a few wet lab work and were not able to find an answer to this pattern.

We have also took a few samples from the Mortazavi's data and were able to replicate the same pattern. This is telling us that for some reason, when looking at the entire RNA Seq run, we see a higher number of reads mapping to short transcript and lower number of reads mapping to longer transcript. The question is why? You expect to see similar averages, especially when you know that you are using the random primed protocol.

Do you see the same bias towards short transcripts in your RNA SEQ runs?
Thanks in advance.
Victor

Zigster · 01-06-2010, 09:31 AM

The most highly expressed genes are short. This is presumably because there is a strong evolutionary pressure for transcriptional efficiency.

vruotti · 01-06-2010, 09:46 AM

Hi Zigster,
You might be right. But, we don't see the same pattern when looking at microarray expression in human. The average expression is about the same for short and long transcripts.

Victor

Zigster · 01-06-2010, 10:17 AM

I think measuring absolute expression in microarrays is very difficult or impossible at the top end of the spectrum. Basically microarrays are only good for ratios, if that.
http://www.pnas.org/content/99/11/7554.full

This article might shed some light on the subject (although I am unable to access it)
Relationship between gene compactness and base composition in rice and human genome. Journal of biomolecular structure & dynamics 2010;27(4):477-88

http://www.jbsdonline.com/c4297/Rela...88-p17743.html

This subject has interested me since we used to see high viral integration in the very shortest genes, which made for some wacky u-shaped graphs. I'd like to hear more about what you discover.

Zigster · 01-06-2010, 11:09 AM

a better article:
Selection for short introns in highly expressed genes
http://www.nature.com/ng/journal/v31/n4/abs/ng940.html

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
paired-end RNA-Seq of short transcripts	fbarreto	Illumina/Solexa	3	11-23-2011 12:43 PM
PubMed: Nucleotide Bias Observed with a Short SELEX RNA Aptamer Library.	Newsbot!	Literature Watch	0	07-29-2011 03:00 AM
RNA-Seq: Improving RNA-Seq expression estimates by correcting for fragment bias.	Newsbot!	Literature Watch	0	03-18-2011 03:00 AM
RNA-Seq: Length Bias Correction for RNA-seq Data in Gene Set Analyses.	Newsbot!	Literature Watch	0	01-22-2011 03:02 AM
How to use short reads to find transcripts and non-coding RNA	chengeng	Bioinformatics	1	07-31-2009 01:40 AM

01-06-2010, 08:19 AM	#1
vruotti Member Location: US Join Date: Feb 2008 Posts: 13	RNA SEQ Bias toward short transcripts (RPKM) Hi there. Just wanted to post this to see if anyone else see this interesting pattern. When taking the average expression (RPKMs) of short genes, say 0.5kb to 2.5kb, we see that the average is much higher when comparing to the average of longer genes, say 10.5-50.5kb. We have done a few wet lab work and were not able to find an answer to this pattern. We have also took a few samples from the Mortazavi's data and were able to replicate the same pattern. This is telling us that for some reason, when looking at the entire RNA Seq run, we see a higher number of reads mapping to short transcript and lower number of reads mapping to longer transcript. The question is why? You expect to see similar averages, especially when you know that you are using the random primed protocol. Do you see the same bias towards short transcripts in your RNA SEQ runs? Thanks in advance. Victor

01-06-2010, 09:31 AM	#2
Zigster (Jeremy Leipzig) Location: Philadelphia, PA Join Date: May 2009 Posts: 116	The most highly expressed genes are short. This is presumably because there is a strong evolutionary pressure for transcriptional efficiency. __________________ -- Jeremy Leipzig Bioinformatics Programmer -- My blog Twitter

01-06-2010, 10:17 AM	#4
Zigster (Jeremy Leipzig) Location: Philadelphia, PA Join Date: May 2009 Posts: 116	I think measuring absolute expression in microarrays is very difficult or impossible at the top end of the spectrum. Basically microarrays are only good for ratios, if that. http://www.pnas.org/content/99/11/7554.full This article might shed some light on the subject (although I am unable to access it) Relationship between gene compactness and base composition in rice and human genome. Journal of biomolecular structure & dynamics 2010;27(4):477-88 http://www.jbsdonline.com/c4297/Rela...88-p17743.html This subject has interested me since we used to see high viral integration in the very shortest genes, which made for some wacky u-shaped graphs. I'd like to hear more about what you discover. __________________ -- Jeremy Leipzig Bioinformatics Programmer -- My blog Twitter

01-06-2010, 11:09 AM	#5
Zigster (Jeremy Leipzig) Location: Philadelphia, PA Join Date: May 2009 Posts: 116	a better article: Selection for short introns in highly expressed genes http://www.nature.com/ng/journal/v31/n4/abs/ng940.html __________________ -- Jeremy Leipzig Bioinformatics Programmer -- My blog Twitter

01-06-2010, 09:46 AM	#3
vruotti Member Location: US Join Date: Feb 2008 Posts: 13	Hi Zigster, You might be right. But, we don't see the same pattern when looking at microarray expression in human. The average expression is about the same for short and long transcripts. Victor