Dear Users,

We have a CyTOF experiment containing 91 samples distributed in 10 batches, with one anchor for normalization in each of these batches.

However, I can't figure out how to make it work: I would like to tell the function which files is in which batches, and which ones are the anchors so that I can use the quantile method, which seems the most appropriate for my needs.

I'm using ncdfFlowSet to import my files in batch.x:

> batch.x
An ncdfFlowSet with 91 samples.
NCDF file : a.nc 
An object of class 'AnnotatedDataFrame'
  rowNames: EC01_Batch02.fcs EC02_Batch02.fcs ... YC23_Batch11.fcs (91 total)
  varLabels: name
  varMetadata: labelDescription

  column names:
    Time, Event_length, Y89Di, Pd102Di, Pd104Di, Pd105Di, Pd106Di, Pd108Di, Pd110Di, Ce140Di, Pr141Di, Nd142Di, Nd143Di, Nd144Di, Nd145Di, Nd146Di, Sm147Di, Nd148Di, Sm149Di, Nd150Di, Eu151Di, Sm152Di, Eu153Di, Sm154Di, Gd155Di, Gd156Di, Gd158Di, Tb159Di, Gd160Di, Dy161Di, Dy162Di, Dy163Di, Dy164Di, Ho165Di, Er166Di, Er167Di, Er168Di, Tm169Di, Er170Di, Yb171Di, Yb172Di, Yb173Di, Yb174Di, Lu175Di, Yb176Di, BCKG190Di, Ir191Di, Ir193Di, Pt195Di, Bi209Di, Center, Offset, Width, Residual

My batch.comp is a list of factors, where one factor contain the names of all files from the same batch, and one factor is used to regroup all the references samples

> batch.comp
$Batch02
[1] EC01_Batch02  EC02_Batch02  FHE01_Batch02 FHE02_Batch02 FHY01_Batch02 YC01_Batch02  YC02_Batch02 
7 Levels: EC01_Batch02 EC02_Batch02 FHE01_Batch02 FHE02_Batch02 FHY01_Batch02 ... YC02_Batch02

$Batch03
[1] EC03_Batch03  EC04_Batch03  FHE03_Batch03 FHY02_Batch03 FHY03_Batch03 YC03_Batch03 
Levels: EC03_Batch03 EC04_Batch03 FHE03_Batch03 FHY02_Batch03 FHY03_Batch03 YC03_Batch03

$Batch04
[1] EC05_Batch04 EC06_Batch04 YC05_Batch04 YC06_Batch04
Levels: EC05_Batch04 EC06_Batch04 YC05_Batch04 YC06_Batch04

$Batch05
 [1] EC07_Batch05  EC08_Batch05  EC09_Batch05  FHE04_Batch05 FHE08_Batch05 FHE09_Batch05 FHE10_Batch05 FHY07_Batch05 YC07_Batch05  YC08_Batch05 
[11] YC09_Batch05 
11 Levels: EC07_Batch05 EC08_Batch05 EC09_Batch05 FHE04_Batch05 FHE08_Batch05 ... YC09_Batch05

$Batch06
 [1] EC10_Batch06  EC11_Batch06  EC12_Batch06  EC13_Batch06  FHE07_Batch06 FHE11_Batch06 FHY08_Batch06 FHY09_Batch06 FHY10_Batch06 YC04_Batch06 
[11] YC10_Batch06 
11 Levels: EC10_Batch06 EC11_Batch06 EC12_Batch06 EC13_Batch06 FHE07_Batch06 ... YC10_Batch06

$Batch07
 [1] EC14_Batch07  EC15_Batch07  EC16_Batch07  FHE12_Batch07 FHE14_Batch07 FHE15_Batch07 FHY12_Batch07 FHY13_Batch07 FHY14_Batch07 FHY15_Batch07
[11] YC11_Batch07  YC12_Batch07 
12 Levels: EC14_Batch07 EC15_Batch07 EC16_Batch07 FHE12_Batch07 FHE14_Batch07 ... YC12_Batch07

$Batch08
 [1] EC17_Batch08  EC18_Batch08  EC19_Batch08  FHE16_Batch08 FHE20_Batch08 FHY16_Batch08 FHY17_Batch08 YC13_Batch08  YC14_Batch08  YC15_Batch08 
10 Levels: EC17_Batch08 EC18_Batch08 EC19_Batch08 FHE16_Batch08 FHE20_Batch08 ... YC15_Batch08

$Batch09
 [1] EC20_Batch09  EC21_Batch09  EC22_Batch09  FHE18_Batch09 FHE21_Batch09 FHE23_Batch09 FHY18_Batch09 FHY19_Batch09 YC16_Batch09  YC17_Batch09 
[11] YC18_Batch09 
11 Levels: EC20_Batch09 EC21_Batch09 EC22_Batch09 FHE18_Batch09 FHE21_Batch09 ... YC18_Batch09

$Batch11
 [1] EC25_Batch11  EC26_Batch11  EC27_Batch11  FHE13_Batch11 FHE24_Batch11 FHE25_Batch11 FHE26_Batch11 FHY22_Batch11 YC22_Batch11  YC23_Batch11 
10 Levels: EC25_Batch11 EC26_Batch11 EC27_Batch11 FHE13_Batch11 FHE24_Batch11 ... YC23_Batch11

$References
[1] Reference_Batch02 Reference_Batch03 Reference_Batch04 Reference_Batch05 Reference_Batch06 Reference_Batch07 Reference_Batch08
[8] Reference_Batch09 Reference_Batch11
9 Levels: Reference_Batch02 Reference_Batch03 Reference_Batch04 Reference_Batch05 ... Reference_Batch11

However, the normalizedBatch function is retourning me a length error.

> normalizeBatch(batch.x, batch.comp, mode="quantile", p=0.05, 
+     target=batch.comp$References, markers=CD14)
Error in normalizeBatch(batch.x, batch.comp, mode = "quantile", p = 0.05,  : 
  length of 'batch.x' and 'batch.comp' must be identical

The length of batch.x is 91 and the length of batch.comp is 10 (the number of batchs plus one for the references).

I read the documentation but I can't figure out how the function is working. I don't have any background in informatics, so I guess that I'm just missing the obvious. Any help would be appreciated!

Best, Florent

> sessionInfo() 
R version 3.5.2 Patched (2019-01-10 r75982)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

locale:
[1] LC_COLLATE=French_France.1252  LC_CTYPE=French_France.1252    LC_MONETARY=French_France.1252 LC_NUMERIC=C                  
[5] LC_TIME=French_France.1252    

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

other attached packages:
 [1] CytoDx_1.2.1                cydar_1.6.1                 SingleCellExperiment_1.4.1  SummarizedExperiment_1.12.0 DelayedArray_0.8.0         
 [6] matrixStats_0.54.0          Biobase_2.42.0              GenomicRanges_1.34.0        GenomeInfoDb_1.18.2         IRanges_2.16.0             
[11] S4Vectors_0.20.1            BiocGenerics_0.28.0         BiocParallel_1.16.6         tidyr_0.8.3                 ggplot2_3.1.0              
[16] FlowSOM_1.14.1              igraph_1.2.4                ncdfFlow_2.28.1             BH_1.69.0-1                 RcppArmadillo_0.9.300.2.0  
[21] flowCore_1.48.1            

loaded via a namespace (and not attached):
 [1] viridis_0.5.1               foreach_1.4.4               viridisLite_0.3.0           ConsensusClusterPlus_1.46.0 shiny_1.2.0                
 [6] assertthat_0.2.1            latticeExtra_0.6-28         GenomeInfoDbData_1.2.0      yaml_2.2.0                  robustbase_0.93-4          
[11] pillar_1.3.1                lattice_0.20-38             glue_1.3.1                  digest_0.6.18               RColorBrewer_1.1-2         
[16] promises_1.0.1              XVector_0.22.0              colorspace_1.4-1            htmltools_0.3.6             httpuv_1.5.0               
[21] Matrix_1.2-17               plyr_1.8.4                  pcaPP_1.9-73                XML_3.98-1.19               pkgconfig_2.0.2            
[26] tsne_0.1-3                  zlibbioc_1.28.0             purrr_0.3.2                 xtable_1.8-3                corpcor_1.6.9              
[31] mvtnorm_1.0-10              scales_1.0.0                later_0.8.0                 tibble_2.1.1                withr_2.1.2                
[36] flowViz_1.46.1              hexbin_1.27.2               lazyeval_0.2.2              magrittr_1.5                crayon_1.3.4               
[41] IDPmisc_1.1.19              mime_0.6                    doParallel_1.0.14           MASS_7.3-51.3               graph_1.60.0               
[46] tools_3.5.2                 rpart.plot_3.0.6            glmnet_2.0-16               munsell_0.5.0               cluster_2.0.7-1            
[51] compiler_3.5.2              rlang_0.3.3                 grid_3.5.2                  RCurl_1.95-4.12             iterators_1.0.10           
[56] BiocNeighbors_1.0.0         bitops_1.0-6                codetools_0.2-16            gtable_0.3.0                rrcov_1.4-7                
[61] R6_2.4.0                    gridExtra_2.3               knitr_1.22                  dplyr_0.8.0.1               KernSmooth_2.23-15         
[66] Rcpp_1.0.1                  rpart_4.1-13                DEoptimR_1.0-8              tidyselect_0.2.5            xfun_0.6          

Both your batch.x and batch.comp are incorrect.

batch.x needs to be a list of ncdfFlowSetobjects. Each element of the list should be one ncdfFlowSet containing all samples for one batch. You're currently supplying a list that contains one ncdfFlowSet object containing samples for all batches. (Alternatively, you can replace each ncdfFlowSet with a list of matrices, with one matrix per sample in that batch; effectively making batch.x a list of lists of matrices.)

batch.comp needs to be a list of the same length as batch.x. Each entry of batch.comp corresponds to the entry at the same position of batch.x. Each entry of batch.comp should contain some identifier specifying the group to which each sample in that batch belongs. This is usually a character vector of length equal to the number of samples in that batch. Each value in that vector indicates the assigned group of a sample in the batch.

Perhaps a demonstration would be more effective. Let's say we have 3 batches:

• Batch 1 contains 9 samples (1 reference, 4 control, 4 treatment A)
• Batch 2 contains 7 samples (1 reference, 3 control, 3 treatment B)
• Batch 3 contains 10 samples (1 reference, 3 control, 3 treatment A, 3 treatment B).

Your batch.comp might then look like this:

batch.comp <- list(
    c("ref", "con", "con", "con", "con", "A", "A", "A", "A"), # batch 1
    c("ref", "con", "con", "con", "B", "B", "B"), # batch 2
    c("ref", "con", "con", "con", "A", "A", "A", "B", "B", "B") # batch 3
)

Obviously, the order of the above elements matters; the ordering of group identities within each element of batch.comp should be determined by the ordering of samples within each element of batch.x. This is analogous to specifying a design matrix in limma or edgeR; you have to get your sample identities right.