Illumina to CNVineta java application

[Mirian Sánchez]

Hello;

 

My name is Mirian. I'm interesting on using the CNVineta software that you have developed with the data of a genome wide genotyping study. I have exported the data from the GenomeStudio as you say in the tutorial (manifest data, full data table and sample table) and now I'm trying to convert those data to a CNVineta format using the "Illumina to CNVineta java application" provided at your webpage.However, I have a problem with this application because when the program stores 190000 markers it blocks. I'm wondering if it is needed a computer with specific characteristics (such as bigger memory size or processor...) or if it is a problem of the application.

 

Thanks in advance.

 

Mirian 

---

Mirian  

---

 

[Mirian]

[Michael]

Dear Miriam,

thank you for reporting that issue.

Using the java application for conversion is unfortunately not so
memory efficient. I've tested the example from our website again (550k
array) and saw that it allocated 1.1 GB RAM. So if you have a computer
with sufficient memory you should try to the following command:
java -jar -Xms512m -Xmx2048m Illu2CNVineta.jar

The -Xms512m parameter tells the java engine to start the application
with 512 mega byte RAM allocated. The parameter -Xmx2048m allows for a
maximum allocation of 2048 mega byte (2 giga byte). Feel free to
change these values if necessary ...
(if this does not work, you could give me access to the manifest file
you use (e.g. via ftp))

The advantage of the java application is the easy handling. But it has
disadvantages like memory efficiency and runtime.

If the solution described above does not work for you, another
suggestion is to convert the Illumina data in analogy to the
Affymetrix Power Tools or QuantiSNP example. This would run faster and
with less memory. If you want to do so, you have to export the LRR and
BAF entries sample wise (one file per sample). The predicted copy
number segments have to be exported sample wise, too. Export the CNV
start position, end position, chromosme and copy number state of the
predicted segments. Eventually you should have a manifest file and two
files per sample (raw data and predicted copy number segments). The
downstream data processing is in analogy to the APT and/or QuantiSNP
conversion.

I hope this helps, if not, do not hesitate to contact us.

Best wishes,
Michael