Re: [aroma.affymetrix] Re: CbsModel

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re: [aroma.affymetrix] Re: CbsModel

Henrik Bengtsson
Hi.

On Wed, Sep 22, 2010 at 9:35 AM, Kai <[hidden email]> wrote:

> Hi Henrik,
>
> Thank you for letting me know the hidden trick. It seems to work. I
> have another related question regarding the CbsModel function.
>
> You've mentioned in addressing another post that for non-affy
> platforms, once one has setup a platform-independent data set (*.asb
> files) as in "Vignette: Creating binary data files containing copy
> number estimates", e.g.
>
> ds <- AromaUnitTotalCnBinarySet$byName(dataSet, tags=tags,
> chipType="*");
>
> One can then pass this to CbsModel just as one passes an
> CnChipEffectSet 'ces' in other vignettes for affymetrix genotyping
> platforms. However, it seems to me that what are stored in the
> CnChipEffectSet are raw CN estimates, whereas the
> "AromaUnitTotalCnBinaryFile" objects contain log2 CN ratios.
>
> If these are correct, my question is that how CbsModel can tell
> whether the input data are in log2-scale or not, or whether the input
> data are ratios or not? Thank you very much for your help on this.

It does this by looking for special tags of the *.asb file.  More
precisely, if the filename has a "log2ratio" tag, then it's content is
assumed to log2-ratio.  Likewise, if there is a "log10ratio" tag, it's
content is assumed to be log10-ratios.  For historical reasons, a
"logRatio" tag is interpreted as "log10ratio".  If none of these tags
exist, the content is assumed to be on the non-logarithmic scale.  I
recommend to use non-logged storage, because that is well defined also
for non-positive values.

DETAILS: The above is taken care of by the AromaUnitTotalCnBinaryFile
class and more precisely the internal/private getAM() method.  The
CbsModel and likewise does not know about this layer and happily
receives log2 ratios regardless of what is stored on file.

Hope this helps

/Henrik

>
> Best,
> Kai
>
>
>
> On Sep 20, 1:06 pm, Henrik Bengtsson <[hidden email]> wrote:
>> Hi Kai.
>>
>> I am aware of the issue, and it is on the todo list to add argument
>> specify that you don't want ratios to be calculated.  There is
>> currently a secret workaround for this that should not be read as an
>> official documented feature [that's a warning for users reading this
>> thread in the future], but it should solve your immediate needs.
>>
>> cbs <- CbsModel(ds);
>> cbs$.calculateRatios <- FALSE;
>>
>> See if that does it for you.
>>
>> /Henrik
>>
>> On Wed, Sep 15, 2010 at 10:14 PM, Kai <[hidden email]> wrote:
>> > Dear Henrik,
>>
>> > I was trying to run CBS model on a set of paired CN estimates. The
>> > data were generated using an Illumina platform, so I have followed
>> > "Vignette: Creating binary data files containing copy number
>> > estimates" to create the log2ratio CN estimates between a tumor sample
>> > and its matched normal.
>>
>> > I have loaded the data with the following codes:
>>
>> > dataSet = "Dataset,tagA,tagB";
>> > chipType = "HumanOmni1-Quad";
>> > ds = AromaUnitTotalCnBinarySet$byName(dataSet,chipType=chipType);
>> > cbs = CbsModel(ds);
>>
>> > However, when I looked at how the CBS model was set up, it says:
>>
>> >> cbs
>> > CbsModel:
>> > Name: Dataset
>> > Tags: tagA,tagB
>> > Chip type (virtual): HumanOmni1-Quad
>> > Path: cbsData/Dataset,tagA,tagB/HumanOmni1-Quad
>> > Number of chip types: 1
>> > Sample & reference file pairs:
>> > Chip type #1 of 1 ('HumanOmni1-Quad'):
>> > Sample data set:
>> > AromaUnitTotalCnBinarySet:
>> > Name: Dataset
>> > Tags: tagA,tagB
>> > Full name: Dataset,tagA,tagB
>> > Number of files: 10
>> > Names: sample1, sample2, ..., sample10 [10]
>> > Path (to the first file): rawCnData/Dataset,tagA,tagB/HumanOmni1-Quad
>> > Total file size: 43.46 MB
>> > RAM: 0.02MB
>> > Reference data set/file:
>> > <average across arrays>
>> > RAM: 0.00MB
>>
>> > It seems to me that the CBS model is using "<average across arrays>"
>> > as reference, which would not be what I want, since my CN estimates
>> > have already been referenced. So my questions are:
>>
>> > 1. Is this how CBS will behave?
>> > 2. Is there a way to let CBS take the CN estimates "as is", without
>> > contrasting to any reference?
>>
>> > Thank you very much for your help on this.
>>
>> > Best,
>> > Kai
>>
>> > --
>> > When reporting problems on aroma.affymetrix, make sure 1) to run the latest
>> > version of the package, 2) to report the output of sessionInfo() and
>> > traceback(), and 3) to post a complete code example.
>>
>> > You received this message because you are subscribed to the Google Groups
>> > "aroma.affymetrix" group with websitehttp://www.aroma-project.org/.
>> > To post to this group, send email to [hidden email]
>> > To unsubscribe and other options, go tohttp://www.aroma-project.org/forum/
>
> --
> When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.
>
>
> You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
> To post to this group, send email to [hidden email]
> To unsubscribe and other options, go to http://www.aroma-project.org/forum/
>

--
When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example.


You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/.
To post to this group, send email to [hidden email]
To unsubscribe and other options, go to http://www.aroma-project.org/forum/