Cnv Calling Methods For The Illumina Omni2.5 - Penncnv And Quantisnp Possible?

Question

[Ryan D]

We would like to use PennCNV and QuantiSNP to call CNVs for new Illumina Omni2.5 array data. These are our normal go-to algorithms for CNV calling. We would like to A) find a way to continue using them or B) find the next best CNV calling algorithm for these arrays.

Problem is PennCNV requires a PFB and HMM file to do CNV calling for these. A PFB file can be generated as Neil showed here. I wondered about the HMM file for the Illumina Omni2.5. I cannot find this anywhere. And I don't know if there is a way to generate it.

For QuantiSNP, it needs an appropriate parameters file and config file as shown here though it is no longer under active development.

I wonder if anyone has solutions that permit either PennCNV or QuantiSNP to be used with the larger Omni2.5 data or, if not, what is the best alternative?

Thanks,

Ryan

[swrdyani]

I want to run penncnv using Perl on win, I got error message penncnv>perl ./detectcnv.pl PennCNV compilation error: Can't locate loadable object for module khmm in @INC (@INC contains: ./ ./kext C:/Perl/site/lib C:/Perl/lib .) at kext/khmm.pm line 8 . Compilation failed in require at ./detectcnv.pl line 10.

PennCNV compilation error: Your system architecture is 'MSWin32-x86-multi-thread ', which is not compatible with pre-compiled executables in PennCNV package. PennCNV compilation error: Please download source code from http://www.openbioin formatics.org/penncnv and compile executable program.

what does that mean

[Robert Sicko]

Did you install Perl to "C:/Perl"? Are you running in CMD or cygwin or other?

[nikmanadli]

hi. i got the same error message, and i ran it in cygwin. can someone please help me rectify this error? I am new in PennCNV, I could not even load the detect_cnv.pl directory, so I am pretty much stuck at the beginning. sigh.

Answer 1

[Ryan D]

So after emailing Christopher Yau, the creator of QuantiSNP, he referred me to someone using QuantiSNP on the new arrays. That individual suggested that the default files are OK. Instead of levels.dat, it is recommended to use the levels-hd.dat and leave params.dat as it is. So yes, it is possible. As for whether it is the best solution or not, we'll find out when we compare the results to our other algorithms.

After running, it looks like QuantiSNP doesn't do well on the Omni data. No regions of CN=0 are reported. PennCNV seems to have more reasonable numbers.

Answer 2

[aatu18]

Hello,

We have genotyped on HumanOmni2.5-8v1_A array and are wondering about pfb and gc files as well. Did you create these files using "all" your signal intensity files? What population, and how many individuals were included for the pfb calculations?

Thanks.

[stevenlang123]

Hello, did you ever find an answer to this?

Answer 3

[Kathleen]

Ryan, have you generated your own pfb file?

[Ryan D]

UPDATED: PFB file is located here: https://dl.dropboxusercontent.com/u/9445847/omni2.5.pfb . Since this PFB was generated from Caucasian samples, you might wish to generate your own.

Answer 4

[Christine]

Hi Ryan,

I noticed that your dropbox for the pfb file is no longer active. I am currently using the Omni2.5 arrays and our lab uses PennCNV as well. I was wondering if you would be able to provide me with the pfb file for these arrays, as well as the GC model file if you have it?

Thanks so much!

Christine

Answer 5

[swrdyani]

Hello there,

I am using QuantiSNP to detect CNVs from the Omni1 quadV1, I run the analyses for all my samples individually and got the row CNV data, do I need to implement the QC inclusion/ exclusion criteria manually, and can anybody walk me through the overlapping of CNV within an individual, between samples, and is PennCNV comparing CNVs from PennCNV, CNVpartition and QuantiSNP ( how can I do so), I would appreciate your quick response, thanks in advance for the help.

[Chris Miller]

Please ask a new question, rather than responding to one that is nearly 2 years old. Feel free to link back to this question if it provides pertinent info.