[BioC] scale of mas5, rma, dchip, and plier results
Rafael A. Irizarry
ririzarr at jhsph.edu
Thu Aug 24 22:56:09 CEST 2006
rma-log2
mas5-original
dchip-original (i think)
as a general rule if the scale is around 0 to 16 its log2 if
its around 0 to 40000 its original scale.
On Thu, 24 Aug 2006, He, Yiwen (NIH/CIT) [C] wrote:
> Hi,
>
>
>
> I'm trying to compare the probe level data analysis tools including
> mas5, rma, dchip, and plier. Using the affy package and plier package, I
> was testing on the affybatch.example data.
>
>
>
> Here's my code:
>
>
>
>> sessionInfo()
>
> Version 2.3.0 (2006-04-24)
>
> i386-pc-mingw32
>
>
>
> attached base packages:
>
> [1] "tools" "methods" "stats" "graphics" "grDevices" "utils"
> "datasets"
>
> [8] "base"
>
>
>
> other attached packages:
>
> plier affy affyio Biobase
>
> "1.4.0" "1.10.0" "1.0.0" "1.10.0"
>
>
>
>> data(affybatch.example)
>
>
>
> #########################
>
> # MAS5:
>
> #########################
>
>
>
>> esetMAS <- mas5(affybatch.example)
>
> background correction: mas
>
> PM/MM correction : mas
>
> expression values: mas
>
> background correcting...done.
>
> 150 ids to be processed
>
>
>
>> dataMAS <- exprs(esetMAS)
>
>
>
>> dataMAS[1:10,]
>
> 20A 20B 10A
>
> A28102_at 15.46299 26.674671 12.68777
>
> AB000114_at 14.95012 32.449716 15.00104
>
> AB000115_at 14.86081 23.366739 12.84647
>
> AB000220_at 13.06876 16.581650 14.25944
>
> AB002314_at 13.25265 7.556996 13.39463
>
> AB002315_at 12.84931 8.891005 12.54743
>
> AB002318_at 15.81936 16.666234 16.58733
>
> AB002365_at 24.55800 22.789565 14.79018
>
> AB002366_at 318.06190 14.877511 19.96851
>
> AC000099_at 13.77299 15.014392 13.15784
>
>
>
> #########################
>
> # RMA:
>
> #########################
>
>
>
>> esetRMA <- rma(affybatch.example)
>
> Background correcting
>
> Normalizing
>
> Calculating Expression
>
>
>
>> dataRMA <- exprs(esetRMA)
>
>
>
>> dataRMA[1:10,]
>
> 20A 20B 10A
>
> A28102_at 4.619839 5.429170 4.456409
>
> AB000114_at 4.529760 6.005186 4.673793
>
> AB000115_at 4.518184 4.949890 4.654417
>
> AB000220_at 4.422690 4.450132 4.422690
>
> AB002314_at 4.335935 3.950547 4.488247
>
> AB002315_at 4.235200 3.842533 4.312301
>
> AB002318_at 4.494510 4.494510 4.494510
>
> AB002365_at 4.679239 4.421983 4.421983
>
> AB002366_at 4.526551 3.919298 4.243068
>
> AC000099_at 4.308961 4.308961 4.221789
>
>
>
> #########################
>
> # dChip:
>
> #########################
>
>
>
>> esetDchip <- expresso(affybatch.example,
> normalize.method="invariantset", bg.correct=F,
> pmcorrect.method="pmonly", summary.method="liwong")
>
>
>
>> dataDchip <- exprs(esetDchip)
>
>
>
>> dataDchip[1:10,]
>
> 20A 20B 10A
>
> A28102_at 101.81628 133.90870 99.43829
>
> AB000114_at 100.92215 145.67048 106.54000
>
> AB000115_at 94.79908 127.11273 95.28790
>
> AB000220_at 97.58051 103.37127 100.09415
>
> AB002314_at 99.53569 83.58326 104.83590
>
> AB002315_at 97.15775 86.10468 101.93582
>
> AB002318_at 101.65877 99.06988 100.42416
>
> AB002365_at 107.30458 99.97733 97.87809
>
> AB002366_at 105.38395 90.52626 98.17004
>
> AC000099_at 112.49714 123.53361 106.67597
>
>
>
> #########################
>
> # PLIER:
>
> #########################
>
>> esetPLIER <- justPlier(affybatch.example)
>
>
>
>> dim(exprs(esetPLIER))
>
> [1] 150 3
>
>
>
>> dataPLIER <- exprs(esetPLIER)
>
>
>
>> dataPLIER[1:10,]
>
> 20A 20B 10A
>
> A28102_at -11.486192 -11.479329 -11.784546
>
> AB000114_at -11.460412 -11.496415 -11.444121
>
> AB000115_at -11.772116 -11.576864 -11.598807
>
> AB000220_at -10.877553 -11.411278 -10.707376
>
> AB002314_at -11.243266 -10.853202 -10.849639
>
> AB002315_at -11.240642 -10.892680 -10.857467
>
> AB002318_at 3.352295 1.780319 1.079215
>
> AB002365_at 3.533914 1.561504 2.719863
>
> AB002366_at 3.972462 -7.970215 1.086926
>
> AC000099_at 4.311271 4.325806 3.339701
>
>
>
> It seems that the expression values from 4 methods are on very different
> scales. Based on a previous message at
> https://stat.ethz.ch/pipermail/bioconductor/2005-May/009146.html,
> justPlier seems to have converted the data to log2. I'm wondering about
> the other methods - mainly, are they in log space or linear space? Do
> people typically see such big difference in the expression values, or is
> it something specific to this sample data?
>
>
>
> Can anyone please help clarifying this?
>
>
>
> Thank you!
>
>
>
> Yiwen He
>
> Contractor, SRA International Inc.
>
> Bioinformatics and Molecular Analysis Section
>
> Center for Information Technology
>
> National Institute of Health
>
> --------
>
> Bldg. 12A, Rm. 1018
>
> 12 South Dr., Bethesda, MD 20892
>
> Email: heyiwen at mail.nih.gov
> Phone: 301-402-4636
>
> Fax: 301-402-2867
>
>
>
>
> [[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
>
More information about the Bioconductor
mailing list