Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2009-01-12 Thread Mark Robinson
Hi Andy. I don't think you've gotten a response on this. Sorry for the delay -- holidays. Some comments below. On 31/12/2008, at 1:18 AM, Andy_Paparountas wrote: > > Hi all , > > I really find this conversation very interesting. I am trying to > analyze a set of 3 treatment and 3 control s

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-30 Thread Andy_Paparountas
Hi all , I really find this conversation very interesting. I am trying to analyze a set of 3 treatment and 3 control samples of MoGeneSt10 array. Thus far with the code pwhite shared I was able to do RMA Background correction , quantile normalization and got QC , RLE , NUSE , density plots. Q1.

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-04 Thread Mark Robinson
Thanks Peter. Perhaps you can repeat this comparison after the next release (this will be very soon!) and split the aroma.affymetrix comparison into: - aroma.affy.oligo -- with RmaPlm(csN,flavor="oligo") - aroma.affy.affyPLM -- with flavor="affyPLM" (as you've done already) Perhaps the best

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-04 Thread pwhiteusa
Dear Mark and Henrik, I wanted to confirm that your summary was correct regarding the different flavors for probeset summarization. I downloaded the MAQC HG_U133_Plus_2 array data from the MAQC website: http://edkb.fda.gov/MAQC/MainStudy/upload/MAQC_AFX_123456_120CELs.zip I then ran the analysis

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-04 Thread pwhiteusa
Hi Mark, I ran your code below on the Mouse Gene ST CDF and agree that like the human array, the probes used by aroma.affymetrix and oligo are identical, appart from the missing 45 control probesets I mentioned previously (I listed them below as requested). For the Human array were the number of

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-04 Thread pwhiteusa
On Dec 3, 8:45 pm, Mark Robinson <[EMAIL PROTECTED]> wrote: > On 04/12/2008, at 10:17 AM, Henrik Bengtsson wrote: > > > So, it all has to do with *how* the log-additive probe-level model is > > *fitted*, correct? > > Correct.  Identical linear model, different procedure for fitting. > > (as a b

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-03 Thread Mark Robinson
On 04/12/2008, at 10:17 AM, Henrik Bengtsson wrote: > So, it all has to do with *how* the log-additive probe-level model is > *fitted*, correct? Correct. Identical linear model, different procedure for fitting. (as a bit of an aside ... I think of these things in terms of influence functio

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-03 Thread Henrik Bengtsson
Hi, thanks Mark for this. So, it all has to do with *how* the log-additive probe-level model is *fitted*, correct? Thus, the model is the same but the algorithms differ. That gives us some sense of how much variance we get from using different algorithms regardless of model. Simulation studi

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-03 Thread Mark Robinson
Hi all. First of all, thanks Peter for 1) doing this testing and 2) for spelling everything out. I expect to refer people to this thread in the future, so thanks for that. Just wanted to add 3 more tidbits of hopefully useful information. 1. I dug a bit into why flavor="oligo" doesn't work

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-03 Thread Henrik Bengtsson
Hi, thanks for sharing all this. On Tue, Dec 2, 2008 at 11:54 AM, pwhiteusa <[EMAIL PROTECTED]> wrote: > > Hi All, > > Here is the exact code I used to analyze Gene ST data for an > experiment performed with the MoGene-1_0-st-v1 array. > > AROMA.AFFYMETRIX > > library(aroma.affymetrix) > cdf <- A

Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-12-02 Thread pwhiteusa
Hi All, Here is the exact code I used to analyze Gene ST data for an experiment performed with the MoGene-1_0-st-v1 array. AROMA.AFFYMETRIX library(aroma.affymetrix) cdf <- AffymetrixCdfFile$fromChipType("MoGene-1_0-st-v1",tags="r3") prefixName <- "Pierson" cs <- AffymetrixCelSet$byName(prefixN

[aroma.affymetrix] Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-11-26 Thread Mark Robinson
Hi all. Just to follow up on these comments here. 'fitPLM' with default parameters in the affyPLM package should give practically identical results to the 'standard' pipeline (RMA bg corr + quantile + fit) within aroma.affymetrix, assuming the underlying annotation is the same. This was a

[aroma.affymetrix] Re: Reproducing RMA with Gene ST data (Was: Re: [aroma.affymetrix] Re: How do you analyze Gene ST Data?)

2008-11-26 Thread Henrik Bengtsson
Hi. a comment on RmaPlm and argument 'flavor': The RmaPlm class is only summarizing the probe signals - normalization etc are done before RmaPlm. The summarization model is the log-additive model with probe affinities and chip effects. The 'flavor' argument specifies which implementation of