SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   RNA Sequencing (http://seqanswers.com/forums/forumdisplay.php?f=26)
-   -   No control for tumour RNA-seq data (http://seqanswers.com/forums/showthread.php?t=40007)

HGENETIC 01-16-2014 07:37 AM

No control for tumour RNA-seq data
 
Hi all,

I'm trying to set up an RNAseq experiment looking at gene expression in brain tumour samples. This is a pilot experiment and we only have 5 tumour samples to work with but I'm not sure what I can use as a control for the experiment? The tumours are pituitary in origin but I have no normal pituitary from the patients the only thing I have is blood. Has anyone had to overcome such a problem and if so how was it done? Any comments would be gratefully accepted,

cheers

H

Bukowski 01-16-2014 10:15 AM

In all honesty I would advise you not to proceed until you have clarified the goals of your experiment. You are not in a position where you can draw meaningful information about your samples other than quantifying the transcripts in them, if this is of interest to you, then go ahead. But there's no route for differential expression. Using blood as a control is of no use to you, all you're going to tell is what is different between blood and brain tumour - hardly the biological question you're going to want to ask.

I suppose you could try and find some 'normal' pituitary RNA-Seq data from somewhere else and use that as a control but I can tell you now that the individual variation between your samples (with so few samples) is likely not going to provide you with anything meaningful.

Why don't you tell us why you are doing this experiment and what you hope to find out from it? That might help more.

HGENETIC 01-17-2014 12:52 AM

Thanks Bukowski for your reply. The aim for this experiment is to see if there are specific genes/biological pathways that are massively up regulated in these tumours, some mouse work has identified increased expression of some members of the wnt pathway so we'd like to be able to replicate those findings in the human and also look for more (the mouse work was microarray). There is also the aim to look for potential gene fusions, alt splicing and RNA editing. We know the tumour is unlike any normal tissue and we are expecting the main driver genes/pathways to be massively over expressed and therefore we are not looking at subtle changes and so that is why we are hoping we may be able to get some kind of reference brain transcriptome data to use as a control.
I'm also wondering if we could normalise our data using some housekeeping genes and compare to reference brain datasets? Apparently the pituitary is a bag of growth hormone so I'm not sure if normal pituitary is a good or bad thing to compare against, my gut instinct is that it is a good comparison but I wonder if there is a whole brain transcriptome dataset out there somewhere?
This is also very much a learning process for our group and so we want to play around with some RNAseq data and optimise an analysis pipeline, I should also add we're doing exome on blood and tumour to look for driver somatic mutations in these samples as well.
I hope that gives you a better understanding of the project and any feedback would be much appreciated

Cheers H

jparsons 01-21-2014 09:40 AM

Illumina's human body map includes data for a bunch of tissues (adipose adrenal brain breast colon heart kidney liver lung lymphnode ovary prostate muscle testes thyroid whiteblood) and has data out there.

Some of the more brain-related bits may be of use to your study, even though pituitary is not included.

HGENETIC 01-22-2014 12:18 AM

Quote:

Originally Posted by jparsons (Post 130279)
Illumina's human body map includes data for a bunch of tissues (adipose adrenal brain breast colon heart kidney liver lung lymphnode ovary prostate muscle testes thyroid whiteblood) and has data out there.

Some of the more brain-related bits may be of use to your study, even though pituitary is not included.

Thanks for that, really appreciate it!

mbblack 02-03-2014 11:00 AM

Quote:

Originally Posted by HGENETIC (Post 129911)
Apparently the pituitary is a bag of growth hormone so I'm not sure if normal pituitary is a good or bad thing to compare against, my gut instinct is that it is a good comparison but I wonder if there is a whole brain transcriptome dataset out there somewhere?

Cheers H

For DGE, that is the very crux of your problem. If the intent is to characterize genes that are differentially expressed in abnormal pituitary tissue, then really, the only valid baseline comparison is against normal pituitary tissue. And if pituitary tissue is normally highly biologically active, then not only may "normal" gene expression be wildly different from other brain tissues, it may also be wildly variable, so the need for large scale replication is essential in order to have any statistical power to discriminate truly differentially expressed genes in the actual tumor samples.

Differentially expression is a purely subjective condition, and as such you absolutely have to properly characterize what you are considering your target tissue against as the baseline. Otherwise, you leave your results entirely open to valid outright dismissal as an inadequate comparison to even address differential expression.

I'm not trying to make your life harder, but seriously, if you do not properly characterize a biologically relevant and legitimate baseline for determination of DGE in your target tissue, you will have nothing that will pass peer review for publication (at the least), and nothing really of any value to anyone. So I would say you absolutely have to get hold of some "normal" pituitary tissue for your controls. At least for the differential gene expression aspect, as the only way to determine if a gene in a cancerous pituitary cell is up or down regulated is to compare it to a normal pituitary cell. Any other comparison is simply invalid for determining DGE in cancerous pituitary cells.


All times are GMT -8. The time now is 07:22 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2020, vBulletin Solutions, Inc.