[BioC] Dye bias adjustment of Illumina Infinium Methylation data

Wed Sep 8 03:34:54 CEST 2010

Hi Tao

The dye bias in the same batch is not a big problem，but dye bias may cause
severe batch effects. I will provide some example data in the lumi package
to show the effects of dye bias when combining data in different batches.
The basic idea of color bias adjustment is to normalize two color channels
each other. You will see more details after I upload the functions to Bioc.
The color bias functions may still need improvements. So your feedbacks are
welcome.  

Pan

On 9/7/10 7:10 PM, "Shi, Tao" <shidaxia at yahoo.com> wrote:

> Hi Pan,
> 
> Could you please clarify the "dye bias" you're referring to?   If you're
> referring to the bias between the two channels, I think it's clear to me from
> Illumina's document and the recent review article from Peter Laird in Nature
> Review Genetics, that there is no need of adjustment, as the two channels in
> Infinium are not corresponding to the methylation states.  The normalization
> method of adjusting median of the two channels as used in methylumi is
> inappropriate for Infinium data.
> 
> Could you please elaborate more on the color bias adjustment functions
> implemented in 
> 
> lumi?  I'm really curious to see.
> 
> Thanks!
> 
> ...Tao
> 
> 
> 
> 
> 
> 
> ________________________________
> From: Pan Du <dupan at northwestern.edu>
> To: Ina Hoeschele <inah at vbi.vt.edu>
> Cc: Simon Lin <S-Lin2 at northwestern.edu>; Sean Davis <sdavis2 at mail.nih.gov>;
> bioconductor at stat.math.ethz.ch
> Sent: Tue, September 7, 2010 10:17:14 AM
> Subject: Re: [BioC] Dye bias adjustment of Illumina Infinium Methylation data
> 
> Hi Ina
> 
> Based on our experience, dye bias does exist in most of our datasets, and
> the bias is usually consistent within the same batch. If all your data is in
> the same batch, then usually no color bias adjustment is necessary for
> Illumina Infinium methylation data. This is consistent with the Illumina
> explanation as you described. However, if your data includes several
> different batches, then dye bias adjustment is important if the dye bias is
> quite different across different batches.
> 
> The lumi package will include some color bias adjustment functions by the
> end of this month (before the new release of Bioc). The lumi package will
> also include support of Infinium 450K arrays in the future.
> 
> 
> Pan
> 
> 
> On 9/7/10 11:59 AM, "Ina Hoeschele" <inah at vbi.vt.edu> wrote:
> 
>> Can someone please clarify for me the need for dye bias adjustment for
>> Illumina Infinium methylation data? Below is an 'explanation' from Illumina
>> to
>> justfy why no dye adjustment is needed:
>> "We actually have two different bead types for each locus, where the 3'
>> terminus corresponds to the methylated and unmethylated condition.  A single
>> base extension is performed and the base after the methylation site is what
>> actually determines the signal (red or green) for the locus.  Since the base
>> proximal to the methylation site will always be the same whether the site is
>> methylated or not, your overall signal will always be in one color for a
>> given
>> locus.  This means that no color normalization is necessary.  This is
>> probably
>> more easily explained by taking a look at cartoon in the attached Data
>> Sheet."
>> 
>> Secondly, will methylumi, lumi and beadarray be able to deal with the new
>> Infinium 450K methylation data?
>> 
>> Many thanks ... Ina
>> 
>> 
>> ----- Original Message -----
>> From: "Sean Davis" <sdavis2 at mail.nih.gov>
>> To: "Chao-Jen Wong" <cwon2 at fhcrc.org>
>> Cc: bioconductor at stat.math.ethz.ch
>> Sent: Thursday, June 17, 2010 5:10:29 PM
>> Subject: Re: [BioC] DNA-Methylation, CopyNumber Results
>> 
>> On Thu, Jun 17, 2010 at 4:43 PM, Chao-Jen Wong <cwon2 at fhcrc.org> wrote:
>> 
>>> As for Illumina Infinium methylation data, you can use the methylumi
>>> package to  manipulate the data and perform quality control and
>>> normalization. Subsequently use limma to do differential methylation
>>> analysis.
>>> 
>>> 
>> Just keep in mind that the data from the Illumina arrays are pretty
>> non-normal, so some assumptions that come into play when using a parametric
>> testing method may be in question.
>> 
>> Sean
>> 
>> 
>> 
>>> On 06/16/10 19:22, Noemi Andor wrote:
>>>> Hi,
>>>> 
>>>> I need to parse some raw Methylation data from Illumina (Infinium) and
>>> some copy number results (CGH, high amount of data). I would be grateful for
>>> any useful information like which bioconducter packages to use, alternative
>>> tools, normalization.
>>>> 
>>>> thank's in advance,
>>>> 
>>>> Noemi
>>>> 
>>>> _______________________________________________
>>>> Bioconductor mailing list
>>>> Bioconductor at stat.math.ethz.ch
>>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>>> Search the archives:
>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>>> 
>>> 
>>> 
>>> --
>>> Chao-Jen Wong
>>> Program in Computational Biology
>>> Division of Public Health Sciences
>>> Fred Hutchinson Cancer Research Center
>>> 1100 Fairview Avenue N., M1-B514
>>> PO Box 19024
>>> Seattle, WA 98109
>>> 206.667.4485
>>> cwon2 at fhcrc.org
>>> 
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>> Search the archives:
>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>> 
>> 
>> [[alternative HTML version deleted]]
>> 
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor at stat.math.ethz.ch
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives:
>> http://news.gmane.org/gmane.science.biology.informatics.conductor
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
> 
> 
> 
>       

--
Pan Du, PhD
Research Assistant Professor
Northwestern University Biomedical Informatics Center
750 N. Lake Shore Drive, 11-176
Chicago, IL  60611
Office (312) 503-2360; Fax: (312) 503-5388
dupan (at) northwestern.edu