[BioC] miRNA normalisation

[James W. MacDonald]

Hi Dermot,

On 1/23/2012 10:18 AM, Dermot Morris wrote:
> Hi Jim,
>
> Thanks for reply
>
> After trying various approaches similar to what you suggested I didn't
> think it was easy!
>
> I'm thinking that even though there are probes for other species on chip
> (other than the one of interest) that even ~specific and non specific
> signals from these would still be better than using a small set (~700
> probe sets) for normalisation.
>
> When I do this and plot summarized "bta" probes only they look ok
> normalised for one tissue extract but not for another. Any comments or
> ideas on this ?

Two things;

First, I am not sure what 'OK' means, so I can't really comment, except 
to say that the differences in these plots may be due to the extraction 
and not the normalization.

Second, I think you might be focusing too much on the bta probes. The 
homology of miRNA is pretty high, and in cases of non-homology, the 
differences tend to be near the ends of the miRNA. Because of this, you 
might find that many of the probes for other species are actually 
expressed in your sample.

Best,

Jim


>
> Regards,
> Dermot
>
>
>
>
> -----Original Message-----
> From: James W. MacDonald [mailto:jmacdon at med.umich.edu]
> Sent: 23 January 2012 15:03
> To: Dermot Morris
> Cc: bioconductor at r-project.org
> Subject: Re: [BioC] miRNA normalisation
>
> Hi Dermot,
>
> On 1/23/2012 9:10 AM, Dermot Morris wrote:
>> Dear all,
>>
>>
>>
>> I would like to  extract the probesets corresponding to a single
> species
>> ("bta")  from an Affymetrix
>>
>> miRNA array  to examine the effects of normalisation using a single
>> species compared to using all probes on array.
>>
>>
>>
>> Is there a simple way to subset an affybatch at the probe level prior
> to
>> normalisation ?
> No. You can subset an AffyBatch at the chip level (essentially
> subsetting by column), but subsetting by row has been deprecated for
> years now.
>
> Hypothetically you could go through the AffyBatch and remove all non-bta
>
> probes by hand. You then might have to do something to the cdf package
> as well. I'm not sure without doing a close examination of the code.
>
> You could also hypothetically use the pm() function with LISTRUE = TRUE,
>
> to extract all the probesets into a named list, then normalize all the
> bta probes separately. You could then put the normalized bta probes back
>
> into that list, convert to a matrix that is correctly ordered (see
> showMethods(pm, class = "AffyBatch", includeDefs = TRUE)), and then use
> the pm() function to put it back into the AffyBatch.
>
> An alternative of this would be to extract as a list, then collect the
> row.names() of all the probesets that are bta probes. You could then
> re-extract using pm() without the LISTRUE argument, and use the
> row.names you extracted to remove those rows, normalize the data, and
> then put back into the original matrix. You could then use pm() to stick
>
> the data back into the AffyBatch and go from there.
>
> You could also hypothetically do the bg.correct() step on your
> AffyBatch, then extract the pm values into a list as above, then do the
> normalization and summarization step by hand on the list (the
> summarization step of rma() can be emulated using the medpolish()
> function).
>
> None of those possibilities is simple, however.
>
> Best,
>
> Jim
>
>
>>
>>
>> Best Regards,
>>
>>
>>
>> Dermot
>>
>>
> ------------------------------------------------------------------------
> ------
>> Attention:
>> This e-mail is privileged and confidential. If you are not the
> intended recipient please delete the message and notify the sender. Any
> views or opinions presented are solely those of the author.
>> This email was scanned by Teagasc and has been certified virus free
> with the pattern file currently in use. This however cannot guarantee
> that it does not contain malicious content.
>> Tabhair aire:
>> Ta an r-phost seo faoi phribhleid agus faoi run. Mura tusa an duine a
> bhi beartaithe leis an teachtaireacht seo a fhail, scrios e le do thoil
> agus cuir an seoltoir ar an eolas. Is leis an udar amhain aon dearcai no
> tuairimi a leiritear.
>> Scanadh an r-phost seo le Teagasc agus deimhniodh go raibh se saor o
> vioras leis an bpatrunchomhad ata in usaid faoi lathair. Ni feidir a
> rathu leis seo afach nach bhfuil abhar mailiseach ann.
>> 	[[alternative HTML version deleted]]
>>
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>

-- 
James W. MacDonald, M.S.
Biostatistician
Douglas Lab
University of Michigan
Department of Human Genetics
5912 Buhl
1241 E. Catherine St.
Ann Arbor MI 48109-5618
734-615-7826

**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues