Question: Normalise counts when tumour samples have normal tissue contamination.
gravatar for Deepali
20 months ago by
Deepali 0
Deepali 0 wrote:


We have 4 control lung samples and 4 tumour lung samples of one mammalian species . In process of taking tumour tissues, the normal tissues are also taken (Don't ask, quite difficult process to take tumour sample from lung). We have virus counts in tumour samples (56500, 35440, 175788, 101192) and very few in control samples (49, 64, 116, 38 - this is due to endogenous virus in host genome). 

Since we have normal tissues present in tumour samples, the gene counts in tumour sample may be effected by this normal tissue contamination. I am looking for a way to normalise the HTSeq-counts. One way that came to my mind was using virus counts along with host genes counts to perform DESeq2 or edgeR. But this came out with not much different than using only host counts. 

Is there any other way to normalise these data to remove effect of normal tissues present in tumour samples using control samples? Or any way we can use these virus counts to normalise the host genes? 

Thank you. 

ADD COMMENTlink modified 20 months ago • written 20 months ago by Deepali 0

I don't have any specific advice or idea of what is the comparison of interest and how it could be achieved.

ADD REPLYlink written 20 months ago by Michael Love25k

Thank you Michael. I can understand this is bit bizarre condition. I will just keep on normal normalisation method using both host and virus genes counts together. 

ADD REPLYlink written 20 months ago by Deepali 0

Do you have an estimate of the normal fraction of the tumour samples from a clinician? If the fraction is about the same for all tumour samples, then it's best to do no normalisation. Fold changes would be under-estimated but would be in the correct order.

ADD REPLYlink written 20 months ago by Dario Strbenac1.5k

Thank you Dario. Fraction is not same in all samples.

Yes you are right the fold change are appeared quite low. 

ADD REPLYlink modified 20 months ago • written 20 months ago by Deepali 0
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 89 users visited in the last hour