Copy Number Variation (Cnv) Detection Using Single-End Sequencing Data

Question

[Doctoroots]

Hi , does anyone know a tool for CNV detection that can work with single-end sequencing data?

the current data im working on is one sample (not case-control) whole exome, but i would appreciate any single-end supporting tool information, (even if it is specific for whole genome).

im already aware of these tools: Breakdancer ; GASV ; PEMer

non of them seem to be appropriate.

thanks

[Sean Davis]

Do you have a comparison, i.e. tumor and normal from the same patient?

[Sean Davis]

Whole genome or just exome?

[Doctoroots]

i dont have a comparsion, and the data is whole exome. if there are specific tools regarding each of these settings i would love to know about it.

Answer 1

[Chris Miller]

You need to clarify whether you're looking for Structural Variants (SV) or Copy-number Variants (CNV) or both. Copy-number calls are generally based on read-depth and it shouldn't matter whether your data is paired-end or not. There are several tools that will happily call copy number from single-sample, single end data. (for example readDepth). Tools that you mention like Breakdancer, on the other hand, are looking for structural rearrangement specifically by using discordantly mapping read pairs.

If you're using exome data that's a whole different ballgame, though. You don't have whole-genome coverage and differences in probe affinities make the read depth vary wildly. Thus calling CNVs with single-sample data is challenging at best. To my knowledge, there are no existing tools that do this.

There is one tool I know of that can make fairly good copy-number calls from paired tumor-normal exome sequencing data, and that's VarScan, but you're unlikely to find it useful since you don't have a paired sample.

---

**disclaimer. I wrote the readDepth package and contributed to the VarScan2 paper (currently in review)

[gskbioinfo143]

Hey

In the current scenario is there any Tool to detect CNV accurately for Single end read data

Thanks

Answer 2

[Mdeng]

At the moment I am in a similar position. I got 10 Samples, 5 Normal, 5 Tumor - from the same patients. Solid sequenced, 50BP long, single end reads. Mapped them using BWA, applied a quality filter (phred >= 30) snp and indel calling, annotation and coverage filter. Everything fine. Now we want to look at the CNVs. If this is possible with varscan, how? Which command? Or what is the workflow?

P.S. Is phred >= maybe to stringent? Any experience with single end solid reads?

With best,

Mario

[Chris Miller]

If you have specific questions about your sample set, please drop them in a new question. FWIW, if you download VarScan and run "java -jar VarScan.v2.2.5.jar copyCaller -h", you'll see the options for that tool.