How to amend script to Normalize without clustering

[CNicholasMDA]

Hello,
How do I amend the script in order to perform the normalization without using clustering? I ran the script as follows but received an error.

Script I ran:

model <- CytoNorm.train(files = train_files,
                        labels = train_data$Batch,
                        channels = channels,
                        transformList = NULL,
                        FlowSOM.params = NULL,
                        normMethod.train = QuantileNorm.train,
                        normParams = list(nQ = 101,
                                          goal = "mean"),
                        truncate_max_range = FALSE,
                        seed = 1,
                        verbose = TRUE)

CytoNorm.normalize(model = model,
                   files = validation_files,
                   labels = validation_data$Batch,
                   transformList = NULL,
                   transformList.reverse = NULL,
                   normMethod.normalize = QuantileNorm.normalize,
                   outputDir = "Batch Corrected 11-01-24",
                   prefix = "After Correction w clust ",
                   truncate_max_range = FALSE,
                   clean = TRUE,
                   verbose = TRUE)

Error I received:
Error in CytoNorm.train(files = train_files, labels = train_data$Batch, :
FlowSOM.params should contain the parameter nCells.

thank you in advance!

[SamGG]

Hi. Did you try to specify a SOM grid of a single cell, I.e. x_grid=y_grid=1?

[CNicholasMDA]

I did not. I can try that. Thanks.

[SofieVG]

While the solution proposed by Sam could work, you could also try the QuantileNorm.train and QuantileNorm.normalize functions directly (also from this package), without the CytoNorm wrapper which mainly handles the clustering step.

Ensure the distributions in your reference samples are certainly stable, except for technical differences, as the whole distribution will be shifted to fit.