[BioC] QC Affy 6.0 SNP arrays with crlmm or Oligo?

[Benilton Carvalho]

On Tue, Mar 23, 2010 at 8:00 PM, jeremy wilson
<jeremy.wilson88 at gmail.com> wrote:
> I wanted to ask why is there no QC done similar to that of expression
> arrays? I read that the bg correction is not required for SNP chips as
> the S/N ratio of intensities is high but I do not see any way to check
> if the chips are damaged (finger prints, water droplets and etc) due
> to mishandling by the technicians. Is this check done looking at the
> genotype calls and their confidence in the downstream analysis? How
> can I visually see the artifacts?

Hi Jeremy,

I believe one may need to extract the intensitty matrix for the SNP
data and work with someyhing like arrayQualityMetrics to get QC going.

We have been able to successfully identify problematic arrays by
investigating SNR.

> Can I see the effects of normalization with HapMap reference? (Before
> normalization and after normalization plots).

What exactly do you mean? Something like a density plot showing that
after normalization the densities across samples are the same? This is
possible with the crlmm package, but not straightforward b/c our plan
with the pkg is to provide genotype/cn tools.

> Any plans to do integrate the genotyping and copy number analysis as
> done in the BIrdSuite? The paper
> (http://www.nature.com/ng/journal/v40/n10/abs/ng.237.html) talks about
> the necessity of this integration. It would be great to see a BioC
> package doing this as well if you think its needed..

We are working on something like that (among other things).

b

>
> On Thu, Mar 18, 2010 at 3:38 AM, Benilton Carvalho
> <beniltoncarvalho at gmail.com> wrote:
>> Hi Jeremy,
>>
>> you may want to take a look at:
>>
>> http://www.ncbi.nlm.nih.gov/pubmed/19906825
>> http://www.ncbi.nlm.nih.gov/pubmed/18387188
>>
>> in summary, crlmm has been compared to BRLMM, BRLMM-P and, more
>> recently, to Birdseed.
>>
>> Regarding CNV, we currently don't have anything implemented for small
>> samples in oligo/crlmm. But Henrik already gave you some
>> suggestions... ;)
>>
>> b
>>
>> On Wed, Mar 17, 2010 at 8:45 PM, jeremy wilson
>> <jeremy.wilson88 at gmail.com> wrote:
>>> Thanks for the reply dear Benilton..
>>>
>>> how can then one proceed with QC'ing SNP arrays using BioC?
>>>
>>> Considering the quality of arrays are good, does the crlmm algorithm
>>> give good call rates with higher accuracy compared to the Birdseed
>>> algorithm from Affy genome console and the BirdSuit from Broad MIT? I
>>> read in one of your papers
>>> (http://biostatistics.oxfordjournals.org/cgi/content/full/8/2/485)
>>> that the crlmm outperforms other competing algorithms. In the article,
>>> was the crlmm alg compared to DM alg or Birdseed? Does crlmm perform
>>> better even now when compared to the Birdseed?
>>>
>>> In my case, how should I do the CNV analysis with only 8 samples. Are
>>> there any other packages or adjustments you can suggest me to
>>> compensate for the small sample size?
>>>
>>> Please let me know
>>> Thank you
>>>
>>> On Wed, Mar 17, 2010 at 12:17 PM, Benilton Carvalho
>>> <beniltoncarvalho at gmail.com> wrote:
>>>> Hi Jeremy,
>>>>
>>>> currently there isn't anything like arrayQualityMetrics implemented in
>>>> oligo/crlmm.
>>>>
>>>> About the sample size, you need at least 10 samples to use
>>>> crlmm::computeCopynumber.
>>>>
>>>> b
>>>>
>>>> On Wed, Mar 17, 2010 at 4:45 PM, jeremy wilson
>>>> <jeremy.wilson88 at gmail.com> wrote:
>>>>> Dear all,
>>>>>
>>>>> I am wondering if there is a QC methodology like the "contrast QC"
>>>>> check by the affymetrix genome console to QC the arrays before
>>>>> genotyping. The crlmm and Oligo packages do the normalization and
>>>>> summarization which is awesome but I do not see it doing QC checks. I
>>>>> am trying to see the quality of the chips with nice plots similar to
>>>>> that of from "arrayQualityMetrics" package for gene expression arrays.
>>>>>
>>>>> One more question: I have only 8 arrays from the same lab and I need
>>>>> to do LOH and copy number analysis. Will the small number of arrays be
>>>>> problematic?
>>>>>
>>>>> Awaiting for your reply,
>>>>> Thank you
>>>>> JW
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>>>>
>>>>
>>>
>>
>