[BioC] marray package: maNorm function with "loes" normalization gives irrreproducable results

[Jean Yee Hwa Yang]

Hi Rudy,

The "global loess" option in maNorm has a sampling component when 
estimating the global loess line.  The default samples 2000 points for 
estimating the overall loess line.  Therefore, you will observed slight 
difference in the estimation from multiple runs.

However, the difference in your data does seems to be much larger than I 
have typically encounter.  This indicates that the estimated global loess 
lines from random subsets of data is very different and you may like to 
investigate the reasons for such difference.  One example, is to use 
"print-tip" loess and compared the difference.

If you like to turn off the sampling option and tell maNorm to use all 
data in the global loess estimation, please let me know and I will send 
you the modified code.

Cheers

Jean

On Thu, 29 Nov 2007, Rudy van Eijsden wrote:

> Hi,
>
>
>
> I noticed that the maNorm function with the "loes" normalization option from
> the marray package gives irreproducible results.
>
> When performing the maNorm function "mnorm <- maNorm(mraw, norm="l")" on the
> same marrayRaw object twice, the results for the M-values differ:
>
>
>
> *Attempt 1:
>
>> mnorm <- maNorm(mraw, norm="l")"
>
>> mnorm at maM[1:10,1:3]
>
>      Data/2006022035.txt Data/2006022036.txt Data/2006022037.txt
>
> [1,]         0.018879167        -0.021035459          0.27322552
>
> [2,]        -0.179815619         0.192413729         -0.10901411
>
> [3,]         1.198385736        -1.086524647         -0.60352712
>
> [4,]        -1.371507828        -0.105926648          1.70023850
>
> [5,]         3.312666352         3.995601174         -0.14438617
>
> [6,]        -0.477144371         0.203647479          0.12004524
>
> [7,]        -0.064336607        -0.264641717          0.27210932
>
> [8,]         0.007128637         0.086065114          0.41770312
>
> [9,]         0.052644302         0.001226107          0.52679449
>
> [10,]        -0.106066947         0.205590655         -0.01910152
>
>
>
> *Attempt 2:
>
>> mnorm <- maNorm(mraw, norm="l")"
>
>> mnorm at maM[1:10,1:3]
>
>      Data/2006022035.txt Data/2006022036.txt Data/2006022037.txt
>
> [1,]          0.03096721          0.03422194          0.31516138
>
> [2,]         -0.22138448          0.19523012         -0.13634274
>
> [3,]          1.00893182         -1.51642222         -0.53571460
>
> [4,]         -1.41207840         -0.12694801          1.67356748
>
> [5,]          2.84049228          3.15626457         -0.02876272
>
> [6,]         -0.47461025          0.20277810          0.09238030
>
> [7,]         -0.14826379         -0.30892746          0.23741928
>
> [8,]         -0.06984206          0.04507657          0.39866073
>
> [9,]          0.06391159          0.07672604          0.56632752
>
> [10,]         -0.14816946          0.20583996         -0.03607932
>
>
>
>
>
> This is de session info:
>
>> sessionInfo()
>
> R version 2.6.0 (2007-10-03)
>
> i386-pc-mingw32
>
>
>
> locale:
>
> LC_COLLATE=English_United States.1252;LC_CTYPE=English_United
> States.1252;LC_MONETARY=English_United
> States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252
>
>
>
> attached base packages:
>
> [1] tools     stats     graphics  grDevices utils     datasets  methods
> base
>
>
>
> other attached packages:
>
> [1] Biobase_1.16.1 marray_1.16.0  limma_2.12.0
>
>
>
>
>
> I have also noticed the same problem on a Linux machine running the same
> version of R and the marray package.
>
>
>
>
>
> Any suggestions?
>
>
>
>
>
> Greetings,
>
>
>
> Rudy.
>
>
>
> =================================================
> Rudy van Eijsden
>
> VIB MicroArray Facility
> Herestraat 49
> 3000 Leuven
> Post Box: 816
> Belgium
>
> e-mail:
> <mailto:Rudy.vanEijsden at vib.be> Rudy.vanEijsden at vib.be
>
> web:
> <http://www.microarrays.be> http://www.microarrays.be
> <blocked::http://www.vib.be/> http://www.vib.be
>
> tel: +32 (0)16 34 79 33
> fax: +32 (0)16 34 79 40
> =================================================
>
>
>
>
> 	[[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
>