Software tools for detecting Copy Number Variations (CNV)

[smice]

Hello,

I'm currently interested in CNV (Copy Number Variation, but I'm sure you're familiar ) detection, and I wonder

1. what software tools exist for this application? I've seen one package so far, found here on SEQanswers: http://tiger.dbs.nus.edu.sg/cnv-seq/ Copyright © 2009 Xie Chao @ Tammi Lab.

2. if other tools can be used for CNV detections. I mean, there are quite a few ChIPSeq analysis tools, which are designed for detecting "peaks": prominent read densities compared to a background read level. The idea behind CNV is basically the same...

(Although I know that some ChIPSeq tools certainly can't be used for CNV, because of their underlying detection concept, like looking for pair of peaks etc.)

Please share your thoughts! Thanks,

Miklos

[bioinfosm]

other tools like SVDetect also provide some copy number information. Do you have any feedback on how CNV-Seq worked for the data?

[Torst]

I'm currently interested in CNV (Copy Number Variation, but I'm sure you're familiar ) detection

I can't recommend any particular software, but one must remember that there is often biases in the sequencing technologies, such that coverage is related to GC content etc, so true CNVs can be masked/missed or exaggerated by these biases. Just keep it in mind.

[bioinfosm]

Torst, thats a good point. And that is why using a normal or baseline sample to compare against is the best bet to overcome sequencing/GC/alignment biases and get truer CNV candidates

[henry.wood]

To add further complication to the GC story, I have found that the GC bias is different between different libraries. Good quality DNA seems to have more of a bias than more degraded samples. This can cause problems if the bias in the test and control sample is different. We've started using a LOESS style correction to account for this.
In answer to the original poster, CNV-seq is a nice program to use for looking at germline variation, but it isn't so good for tumours. I've been using DNAcopy from bioconductor for my tumour samples.

[smice]

other tools like SVDetect also provide some copy number information. Do you have any feedback on how CNV-Seq worked for the data?

You mean, how ChIPSeq worked for the CNV data? No, we haven't done any analyses, currently we are in the experiment designing phase, wondering whether we should develop a CNV detecting tool ourselves or use existing ones (if there are any). And this idea came up to use ChIPSeq peak detection tools for CNV.

I thank all of you very much for your advices! We are aware of the possible obstacles, that's why I'm asking the community whether they know/use some tool that is able to overcome them all.

I share my experiences as soon as we analyzed the CNV data!

[westernblotter]

Can you please elaborate further on why the good quality DNA has greater GC bias than more degraded DNA?

[henry.wood]

Can you please elaborate further on why the good quality DNA has greater GC bias than more degraded DNA?

I'm not completely sure. I thought it was something to do with regions that are GC rich having a denser secondary structure in vivo. If the DNA is still very good quality some of this will remain, so the AT rich regions will preferentially shear. If the DNA has been through a few freeze/thaws then the DNA isn't so coiled up and the shearing is more even. That might be gibberish, but it's the best we can come up with.

[krobison]

BTW, catalog of programs in this space.

A new tool FREEC claims to adjust for GC-bias; I have not tried it.

[zlu]

BTW, catalog of programs in this space.

A new tool FREEC claims to adjust for GC-bias; I have not tried it.

Before trying FREEC, I have a question which may have a very simple answer. If I have a diploid organism and FREEC predicts the sample to have a gain of 6 copies; am I right to say that there are 3 copies each on the 2 alleles or is the gain refers to 6 copies each? Thank you.

[difereg]

Hello All!
I am trying to define the best methodology to study CNV from gDNA parasite sequences.
Can I used CNV-seq in analyses of haploid parasites?
thanks

[valeu]

Control-FREEC is able to find CNVs in all kinds of organisms including haploid parasites

[valeu]

It means that there are 6 copies in total. If you want to know which exactly alleles were duplicated, you need to run FREEC on pileup files and use option [BAF]. But coverage must be rather high (>20x) to be able to detect the genotype.

[gprakhar]

Hello,

CNVnator is one of the best tools for CNV detection, according to my analysis, for a single sample only.
If you have a Tumor normal pair kind of data then FREEC is a good option.

--
Prakhar Gaur

[BLC]

I'd like to add our flagship product - Nexus Copy Number. We can debate the 'best' tool for CNV analysis, so I won't get into that -- but I would like to provide a list of references for where the product has been cited in peer-reviewed publications. (http://www.biodiscovery.com/corporate/publications/).

I will say the key features are entirely user-interface driven (no programming/scripting); very high performance for handling hundreds or even thousands of samples together on a desktop machine; and genomic platform independent. There are too many 'features' and listing them out here wouldn't be appropriate.

We provide fee trials of the software which you can use to analyze your own data, and our goal is to maintain a state-of-the-art with respect to advances in the field.