Greetings.

Using wateRmelon (1.6.0)  I ran into the following error 

Error in betaqn(bn = getBeta(object)) : 
  error in evaluating the argument 'bn' in selecting a method for function 'betaqn': Error in getGreen(rgSet)[getProbeInfo(rgSet, type = "II")$AddressA, ] : 
  subscript out of bounds

with a data set for which this previously worked. This is after a pfilter step which removed 1 sample and  6239 sites.

The betaqn function works on the object not subjected to pfilter suggesting that pfilter is producing an object that is not quite aligned. Possibly the list of Green readings is not correctly updated  (not obvious since the object is in the environment and the relative parts are not 'exposed').

As noted in one of my answer to myself:

I did some further digging and it turns out the error is with getProbeInfo (package minfi). It is returning the info for the manifest  regardless of the FeatureSet object. Thus the returned subscript vector is always longer than the dimensions of the reduced matrix (after filtering).

The function getGreen returns a matrix from the pfilter object that has fewer rows than the vector returned by getProbeInfomData.pf, type = "II")$AddressA.

Now to see when this behavior changed.

Gerard

PS. have sent the maintainer of minfi an e-mail asking about the getProbeInfo behavior.

Current version and sessioninfo.
================================
R version 3.1.2 (2014-10-31)
Platform: x86_64-redhat-linux-gnu (64-bit)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats4    parallel  stats     graphics  grDevices utils     datasets 
[8] methods   base     

other attached packages:
 [1] IlluminaHumanMethylation450kmanifest_0.4.0        
 [2] sva_3.12.0                                        
 [3] genefilter_1.48.1                                 
 [4] mgcv_1.8-4                                        
 [5] nlme_3.1-119                                      
 [6] wateRmelon_1.6.0                                  
 [7] IlluminaHumanMethylation450kanno.ilmn12.hg19_0.2.1
 [8] ROC_1.42.0                                        
 [9] lumi_2.18.0                                       
[10] methylumi_2.12.0                                  
[11] ggplot2_1.0.0                                     
[12] reshape2_1.4.1                                    
[13] scales_0.2.4                                      
[14] matrixStats_0.13.1                                
[15] limma_3.22.4                                      
[16] minfi_1.12.0                                      
[17] bumphunter_1.6.0                                  
[18] locfit_1.5-9.1                                    
[19] iterators_1.0.7                                   
[20] foreach_1.4.2                                     
[21] Biostrings_2.34.1                                 
[22] XVector_0.6.0                                     
[23] GenomicRanges_1.18.4                              
[24] GenomeInfoDb_1.2.4                                
[25] IRanges_2.0.1                                     
[26] S4Vectors_0.4.0                                   
[27] lattice_0.20-29                                   
[28] Biobase_2.26.0                                    
[29] BiocGenerics_0.12.1                               

loaded via a namespace (and not attached):
 [1] affy_1.44.0             affyio_1.34.0           annotate_1.44.0        
 [4] AnnotationDbi_1.28.1    base64_1.2              base64enc_0.1-2        
 [7] BatchJobs_1.5           BBmisc_1.9              beanplot_1.2           
[10] BiocInstaller_1.16.1    BiocParallel_1.0.3      biomaRt_2.22.0         
[13] bitops_1.0-6            brew_1.0-6              checkmate_1.5.1        
[16] codetools_0.2-10        colorspace_1.2-4        DBI_0.3.1              
[19] digest_0.6.8            doRNG_1.6               fail_1.2               
[22] GenomicAlignments_1.2.1 GenomicFeatures_1.18.3  grid_3.1.2             
[25] gtable_0.1.2            illuminaio_0.8.0        KernSmooth_2.23-13     
[28] MASS_7.3-37             Matrix_1.1-5            mclust_4.4             
[31] multtest_2.22.0         munsell_0.4.2           nleqslv_2.5            
[34] nor1mix_1.2-0           pkgmaker_0.25.8         plyr_1.8.1             
[37] preprocessCore_1.28.0   proto_0.3-10            quadprog_1.5-5         
[40] R.methodsS3_1.6.2       RColorBrewer_1.1-2      Rcpp_0.11.4            
[43] RCurl_1.96-0            registry_0.2            reshape_0.8.5          
[46] rngtools_1.2.4          Rsamtools_1.18.2        RSQLite_1.0.0          
[49] rtracklayer_1.26.2      sendmailR_1.2-1         siggenes_1.40.0        
[52] splines_3.1.2           stringr_0.6.2           survival_2.37-7        
[55] tools_3.1.2             XML_3.98-1.1            xtable_1.8-0           
[58] zlibbioc_1.12.0   

Trying to figure out how to edit my own post. 

This is an error with the pfilter function, not betqn. The object returned by pfilter is the problem. The functions from other packages also generate similar errors on the object returned by pfilter. 

Gerard

Did some further digging and it turns out the error is with getProbeInfo (package minfi). It is returning the info for the manifest  regardless of the FeatureSet object. Thus the returned subscript vector is always longer than the dimensions of the reduced matrix (after filtering).

The function getGreen returns a matrix from the pfilter object that has fewer rows than the vector returned by getProbeInfomData.pf, type = "II")$AddressA.

Now to see when this behavior changed.

PS. First thought getProbeInfo was the function from oligo, but it is exposed by minfi.

Resolved

Looks like the problem is with the minfi object. The object is generated is a RGChannelSetExtended object and stored without a featureData@data component. What I mean is that it is indirect. Therefore the subsetting of the object does not subset the annotation. The normalization methods (regardless of which one is used) all rely on subsetting in a way that expects the annotation to already have been subsetted. That is, the normalization methods do not work on RGChannelSetExtended objects that have probes removed (are a subset of the probes in the annotation).

I have verified that using methylumIDAT() from the methylumi package to read in the data results in a MethyLumiSet object and that behaves as expected for the normalization functions. That is: one can remove probes (pfilter) that do not pass QC and subsequently adjust (normalize) the data.