[BioC] LPE error and warnings

[Stephanie Mewborn]

I'm trying to run LPE on my data set (2 replicates each, 2 samples=4 
chips, hgu133plus2).  Here's what I'm getting:

 > data[1:3,]
          ID E.161K.DCM.A.CEL E.161K.DCM.B.CEL ISC.Cntl.A.CEL 
ISC.Cntl.B.CEL
1 1007_s_at         9.809272        10.158500       9.648349       
9.700356
2   1053_at         7.123087         7.335424       6.911521       
6.811536
3    117_at         7.951087         7.687096       7.724805       
7.792392
 > data[,2:4] <- preprocess(data[,2:4], data.type = "MAS5")
 > data[1:3,]
          ID E.161K.DCM.A.CEL E.161K.DCM.B.CEL ISC.Cntl.A.CEL 
ISC.Cntl.B.CEL
1 1007_s_at         3.412529         3.428885       3.402416       
9.700356
2   1053_at         2.950885         2.995280       2.896051       
6.811536
3    117_at         3.109535         3.049017       3.060164       
7.792392
 > data <- data[substring(data$ID, 54614:54675) !="AFFX",]
 > dim(data)
[1] 54675     5
 > data <- data[substring(data$ID,1,4) !="AFFX",]
 > dim(data)
[1] 54613     5
 > data[1:3,]
          ID E.161K.DCM.A.CEL E.161K.DCM.B.CEL ISC.Cntl.A.CEL 
ISC.Cntl.B.CEL
1 1007_s_at         3.412529         3.428885       3.402416       
9.700356
2   1053_at         2.950885         2.995280       2.896051       
6.811536
3    117_at         3.109535         3.049017       3.060164       
7.792392
 > var.Naive <- baseOlig.error(data[,2:3], q=0.01)
 > dim(var.Naive)
[1] 54613     2
 > var.Naive[1:3,]
          A        var.M
1 3.420707 0.0006387559
2 2.973082 0.0037404778
3 3.079276 0.0035043507
 > var.wt <- baseOlig.error(data[,4:5], q=0.01)
 > dim(var.wt)
[1] 54613     2
 > var.wt[1:3,]
          A    var.M
1 6.551386 58.44331
2 4.853794 58.44331
3 5.426278 58.44331
 > lpe.val <-data.frame(lpe(data[,4:5], data[,2:3], var.wt, var.Naive, 
probe.set.name=data$ID))
 > lpe.val <- round(lpe.val, digits=2)
Error in Math.data.frame(x, digits) : Non-numeric variable in 
dataframe: flag.outlier.x
 > dim (lpe.val)
[1] 54613    15
 > lpe.val[1:3,]
           x.ISC.Cntl.A.CEL x.ISC.Cntl.B.CEL median.1 std.dev.1 
p.outlier.x flag.outlier.x y.E.161K.DCM.A.CEL
1007_s_at         3.402416         9.700356 6.551386  7.644823   
0.5602117              .           3.412529
1053_at           2.896051         6.811536 4.853794  7.644823   
0.7172308              .           2.950885
117_at            3.060164         7.792392 5.426278  7.644823   
0.6615988              .           3.109535
           y.E.161K.DCM.B.CEL median.2  std.dev.2 p.outlier.y 
flag.outlier.y median.diff pooled.std.dev
1007_s_at           3.428885 3.420707 0.02527362   0.6622332            
   .    3.130679       5.405736
1053_at             2.995280 2.973082 0.06115945   0.6078503            
   .    1.880711       5.405879
117_at              3.049017 3.079276 0.05919756   0.4682921            
   .    2.347002       5.405868
             z.stats
1007_s_at 0.5791403
1053_at   0.3479011
117_at    0.4341582
 > fdr.BH <- fdr.adjust(lpe.val, adjp="BH")
 > dim(fdr.BH)
[1] 54613     2
 > round(fdr.BH[1:4, ],2)
       FDR z.real
[1,] 0.85   1.00
[2,] 0.85   1.00
[3,] 0.85   0.99
[4,] 0.85   0.99
 > fdr.2 <- fdr.adjust(lpe.val, adjp="resamp", iterations=2)
iteration number 1 is in progress
iteration number 1 finished
iteration number 2 is in progress
iteration number 2 finished
Computing FDR...
Error in "[<-"(`*tmp*`, j, value = numeric(0)) :
	nothing to replace with
In addition: Warning messages:
1: NaNs produced in: sqrt(2 * var.x0)
2: NaNs produced in: sqrt(2 * var.y0)
3: NaNs produced in: sqrt(2 * var.x0)
4: NaNs produced in: sqrt(2 * var.y0)
 >

A previous post regarding the NaNs warning had a reply recommending the 
preprocessing step which I have performed, without receiving an error 
message.  I did have the error for the rounding step though, and my 
lpe.val results include the p.outlier, and flag.outlier columns that 
are not shown in the vignette.    I am running R 2.0 and Bioconductor 
1.5 on a Mac OSX system.


Stephanie K. Mewborn, Ph.D.
Post-Doctoral Scholar

University of Chicago
Department of Medicine, Section of Cardiology
Phone:  773-702-2684
Fax:  773-702-2681