missing value in beta matrix after normalization wich champ.norm
1
0
Entering edit mode
jfertaj ▴ 30
@jfertaj-8566
Last seen 2.2 years ago
United Kingdom

Hi ,

We have a set of 200 samples that we have analysed with EPIC array. I have load the idat.files using champ.load, with autoimpute=T.  The output has no missing values, but when I normalize using champ.norm to run champ.SVD I get missing value again. is that a normal behaviour? should I impute again the matrix after using champ.norm?

Thanks

here is my code:

myLoad_noXY <- champ.load(directory=getwd(),
                  filterXY = TRUE,
                   SampleCutoff = 0.9,
                   filterDetP=F,
                   filterBeads=F,
                   ProbeCutoff = 0.02,
                   detPcut = 0.01, 
                   beadCutoff = 0.05, 
                   filterMultiHit = F, 
                   filterSNPs = F, 
                   filterNoCG = F,
                   arraytype="EPIC",
                   autoimpute = TRUE,
                   method = "minfi")

# Normalise data
myNorm <- champ.norm(beta=myLoad_noXY$beta,
               rgSet=myLoad_noXY$rgSet,
               mset=myLoad_noXY$mset,
               resultsDir="./CHAMP_Normalization_noXY_imputed/",
               method="BMIQ",
               plotBMIQ=FALSE,
               arraytype="EPIC",
               cores=3)

# SVD
champ.SVD(beta = myNorm,
              rgSet=NULL,
              pd=myLoad_noXY$pd,
              RGEffect=FALSE,
              PDFplot=TRUE,
              Rplot=FALSE,
              resultsDir="./CHAMP_SVDimages_noXY/")
champ champs.svd() champ.norm • 1.5k views
ADD COMMENT
0
Entering edit mode
Yuan Tian ▴ 290
@yuan-tian-13904
Last seen 7 months ago
United Kingdom

Hello:

I don't think normalization step would induce missing value. Actually, missing value is not allowed to exist for data to be normalized. Could you tell me more about your code? 

Best

Yuan Tian

ADD COMMENT
0
Entering edit mode

Tnanks @Yuan Tian, I have pasted my code in the question

ADD REPLY
0
Entering edit mode

Hello:

I just checked the code. I wonder if you have checked after loading, is there missing value in your myLoad$beta variable? In your code, I can see filterDetP=F. I think it means imputation would not be done because NA values are actually induced by detect p value, if filterDetP=F, it could mean imputation is not working.

By the way, when you run champ.load() code, have you see the message like imputation on your screen? BMIQ method used two quantile function projection solution to do normalization, I think if original beta matrix is free of NA or negative value, the normalized result should not induce NA.

Anyway, could you tell me how many NAs in beta matrixes: myLoad$beta, myNorm?

Best

Yuan Tian

ADD REPLY

Login before adding your answer.

Traffic: 424 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6