Sequenza on tumor only samples

[Shobana Sekar]

Hi,

Is it possible to run Sequenza on samples that do not have a matched normal? Is there any work-around for samples that are tumor only?

Thanks!

Shobana

[Francesco Favero]

Hi Shobana,

Short answer, yes….

I’ve managed to have decent performance on sample without a reference, but in that case I was using a random normal WXS to compare the depth.

The genotyping, in real patient data, it is possible to estimate from the tumor sample only, although when the purity is high enough, all the LOH region would suffer a lack of detection of heterozygous positions.

If you don’t have a normal WXS on hand you can identify the more normal-like sample (respect of the CNV profile) and use that as normal.

It is a workaround, and it might not work perfectly, but you can manage to set it up. Are you using varscan or the python script to generate the input files?

Both case it is possible to have the desired result without problem.

Best

Francesco

[Weini Huang]

Hi Francesco, 

I have a related questions. I have a drug-sensitive cell line (ancestor sample) and  drug-resistant cell lines  (evolved samples). Thus I also don't have the matched normal samples. I want to know the difference of CNVs between ancestor and evolved samples. How should I run Sequenza for such samples to get a meaningful results? 

For a test, I used my more normal-like sample - drug sensitive cell line- as normal, and evolved drug-resistant cell line as tumour. The results looks very suspicious to me. For example, the red bars in the genome view of copy number can up to 7 in many chromosomes. Could you give me some suggestions on this?

Thank you in advance,

Weini

[Francesco Favero]

Hi Huang,

I would do the same thing as you did, using the ancestor cell line as reference.

With cell lines the optimal setting might narrow the cellularity solution to 0.9 to 1 (or something similar), and there might be some problem in location where the ancestor (if) have loss of heterozygosity, where the daughter cell line would not have allelic information.

Beside all that, if your bam files where performed in different times, it might present bias such as replication time:

look at this file on IGV (or your preferred viewer)

http://hgdownload.cse.ucsc.edu/goldenPath/hg19/encodeDCC/wgEncodeUwRepliSeq/wgEncodeUwRepliSeqSknshWaveSignalRep1.bigWig

if your results have a profile remarkably similar to this pattern it might be effected by the bias.

I’m working on a solution about the above-mentioned bias (if you need it and if you want, I can share the work in progress solution, although it’s a bit rough at the moment).

Best

Francesco

--
You received this message because you are subscribed to the Google Groups "Sequenza User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sequenza-user-g...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[pin wu]

Hi Francesco,

I encountered the same problem. I have WES data for a batch of tumor samples without control group. I want to use sequenza software to detect copy number variations in tumor samples. Based on your suggestions above, can I choose a random normal sample from other patients of the same cancer as control? Or is it better to combine the sequencing data of normal samples from multiple patients as a control?

Looking forward to your reply.

Best regards,

wupin

[Diana Ivanoiu]

Hi Francesco, 

I have a similar question/issue as someone above. I basically have WGS for the same cell line at different timepoints of treatment. I have used as normal my time 0 and I also get funny results. Also interestingly, I do get low cellularity and CNVs even if I run against this time 0 (normal) another time 0 replicate - tumour. It is basically just another subset of cells taken from the same flask - so I guess one would expect to see nothing? I was using this more of a control but it detects a lot of events. 

Any advice on this analysis? 

Many thanks, 

Diana