Normalization with champ.norm in ChAMP Pakcage

[Heidi Kong]

I am trying to use champ.norm function in the Bioconductor package ChAMP for our DNA methylation data for 389 samples.
When I run champ.norm function I found a problem. At the first time it stopped at the 172th sample, at the second time it stopped at the 53rd sample, at the third time, and the fourth time, and so on, it just stopped at some samples.
Could any one please tell me what is happening here, and what could be a good solution for this?

[Rob]

Could be a memory issue what OS and how much RAM do you have?

[Heidi Kong]

Hi Rob,

  I am using cluster and RAM size is 100GB.

  I set plotBMIQ=FALSE, then it worked fine.

  Do you have an idea why?

  Thank you very much.

On Fri, May 2, 2014 at 4:53 AM, Rob <r.a....@googlemail.com> wrote:

Could be a memory issue what OS and how much RAM do you have?

--
You received this message because you are subscribed to a topic in the Google Groups "Epigenomics forum" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/epigenomicsforum/NLeJ32kQrQA/unsubscribe.
To unsubscribe from this group and all its topics, send an email to epigenomicsfor...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Celine Bourdon]

Hi Rob and Heidi,

I am having the same issue,
the analysis stops at random samples... sometimes getting farther down sometimes stopping very early.
I am running on a linux with 70GB of RAM, and monitoring the memory with htop, and it is not even close to "running out"...

I get the following errors... 
error in density.default(nbeta2.v): 'x' contains missing values

which seems related to making density plots... ? from what i google-gathered...
so turning of plotBMIQ might make sense ...  if my analysis crashes again, this will be my next go-to...

[Morris, Tiffany]

Dear Celine,

This error is caused by zeros in your data when normalising with BMIQ. Are you using the newest version of ChAMP?

Best wishes

Tiffany

-- 

Tiffany Morris, PhD

Postdoctoral Research Associate
Medical Genomics Group,
UCL Cancer Institute,
University College London,
72 Huntley Street,
London, WC1E 6BT

Tel: +44 (0) 20 7679 0999

You received this message because you are subscribed to the Google Groups "Epigenomics forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to epigenomicsfor...@googlegroups.com.

[Celine Bourdon]

Hi Tiffany,

I have ChAMP 1.3.1 ... running with R Spring Dance version 3.1.0  through Rstudio(0.98.507)... on a ubuntu with ~70GB (with a dataset of 329 samples).

Funny that it would be zeros...  because the first step that gets called up seems to be  "replacing zeros with 0.00001 (or some low number of the sort)" ...
and it seems to go through that "phase" fine...?
 if you can let me know how i should treat these zeros (once i find the bad boys).

 (i am also thinking ... I am having problems with this version of R ... some packages cannot be loaded anymore ex: maniString ... used in getManifest, and estimatecellcount of Minfi.... i am looking to revert back to the older R-version..

hope this helps...
celine

[Heidi Kong]

Hi Tiffany,

  Yes, mine is the newest version.

  If I set

      plotBMIQ=FALSE, then it worked fine.

  The problem disappears. I think I don't need the plot.

  What would you recommend?

  Heidi

[Morris, Tiffany]

You don’t need the plot…it is only of interest if you want to see how BMIQ is fitting your data.

[Simon Vary]

Hi,
I ran into same problem today, stopping on 427th sample, while analysing around 800 samples using R 3.1.1 with ChAMP 1.2.7.  It was using a lot of RAM (around 40G) however there was still plenty left.
I disabled plotBMIQ=FALSE, hopefully it will work.
It would be nice to see the fitting too, did anyone figured out what the problem with plotBMIQ=TRUE was ?

Simon

To unsubscribe from this group and all its topics, send an email to epigenomicsfo...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Heidi Kong]

Hi Simon,

  No one could have figured out.

  Sorry.

  Heidi