Issue creating a GatingSet

[santoshe1]

Command -
> fs_transformed A flowSet with 130 experiments.
gs <- GatingSet(fs_transformed)

I get an error:
Error in h(simpleError(msg, call)) : error in evaluating the argument 'x' in selecting a method for function 'GatingSet': unimplemented type 'list' in 'EncodeElement'

Kindly assist. Thanks

[gfinak]

How do you propose we try to reproduce your error?

[mikejiang]

First, try the code below to see if it succeeds

library(flowCore)
library(flowWorkspace)
data("GvHD")
fs <- GvHD[1:2]
gs <- GatingSet(fs)
gs

Then, provide some more context of your use case and code, paste the traceback() output and sessionInfo,
lastly, it would be helpful to provide the minimum reproducible example (as suggested by @gfinak )

[epgui]

@mikejiang the code you posted works for me, no problem.

However, I also am getting the same error as @santoshe1 and this is a (relatively) new problem. I will provide you with a test case that works and a test case that does not work, complete with sample data, tomorrow.

[epgui]

@mikejiang in Experiment 1, a flowset is created successfully but we fail to create a gatingset, with the same error as reported above by @santoshe1.

# > source("https://bioconductor.org/biocLite.R")
# > biocLite()
library("Biobase")
library("flowCore")
library("flowWorkspace")

# Set this to the correct directory
#setwd("/Users/gp/Developer/flowworkspace-test-case")

pData <- read.table("Experiment1_phenoData.txt", row.names=1, header=TRUE, sep="\t")
pData$Filename <- row.names(pData)
row.names(pData) <- pData$Filename
phenoData <- new("AnnotatedDataFrame", data=pData)

flowset <- read.flowSet(
  dataset=1,
  transformation=FALSE,
  alter.names=TRUE,
  phenoData = phenoData
)

gatingset <- GatingSet(flowset)

In Experiment 2, both flowset and gatingset are created successfully.

# > source("https://bioconductor.org/biocLite.R")
# > biocLite()
library("Biobase")
library("flowCore")
library("flowWorkspace")

# Set this to the correct directory
#setwd("/Users/gp/Developer/flowworkspace-test-case")

pData <- read.table("Experiment2_phenoData.txt", header=TRUE, sep="\t")
phenoData <- new("AnnotatedDataFrame", data=pData)

flowset <- read.flowSet(
  dataset=1,
  transformation=FALSE,
  alter.names=TRUE,
  phenoData=phenoData
)

gatingset <- GatingSet(flowset)

Complete test cases: https://drive.google.com/drive/folders/1JJWsCYgMjPeWVFKBm0GfXMp2gTlLDb3I

[epgui]

It should be noted that while I was expecting read.flowSet() to take the row names of the AnnotatedDataFrame as file paths, it actually requires you to have a column specifically identified as "filename" or "file" (not case-sensitive).

Furthermore, it appears that creating an AnnotatedDataFrame in this way fails to use column names as default metadata, such that one must write instead:

pData <- read.table("Experiment2_phenoData.txt", header=TRUE, sep="\t")
metadata <- data.frame(labelDescription=colnames(pData), row.names=colnames(pData))
phenoData <- new("AnnotatedDataFrame", data=pData, varMetadata=metadata)

... to have the metadata fields populated. Otherwise they are filled with NAs.

I understand that this is not an issue with flowWorkspace per se, but if we then take this phenoData object and pass it to read.flowSet(), an additional weirdness happens: the metadata contained in the phenoData is not preserved in the flowset... So I think there may be some funny compatibility issues between biobase and flowCore.

This can all be reproduced with the test case files provided above.

[mikejiang]

It was due to the / characters in sampleNames resulted from the nested subfolders structures in your cases.

head(rownames(pData))
[1] "Experiment1_data/786 n1/786n1_Ca1uM4h_samp_20x 00016119 613.LMD" 
[2] "Experiment1_data/786 n1/786n1_Ca1uM4h_samp_60x 00016122 616.LMD" 
[3] "Experiment1_data/786 n1/786n1_Ca1uM4h_samp_100x 00016125 619.LMD"

This causes the trouble for write.flowSet that is invoked by GatingSet constructor under the hood.
I've pushed the easy fix to flowCore to substitute them with _ within read.flowSet i.e.

flowset <- read.flowSet(
+   dataset=1,
+   transformation=FALSE,
+   alter.names=TRUE,
+   phenoData = phenoData
+ )
> 
> sampleNames(flowset)
 [1] "Experiment1_data_786 n1_786n1_Ca1uM4h_samp_20x 00016119 613.LMD" 
 [2] "Experiment1_data_786 n1_786n1_Ca1uM4h_samp_60x 00016122 616.LMD" 
 [3] "Experiment1_data_786 n1_786n1_Ca1uM4h_samp_100x 00016125 619.LMD"

so now it should work for GatingSet

gatingset <- GatingSet(flowset)
> head(sampleNames(gatingset))
[1] "Experiment1_data_786 n1_786n1_Ca1uM4h_samp_20x 00016119 613.LMD" 
[2] "Experiment1_data_786 n1_786n1_Ca1uM4h_samp_60x 00016122 616.LMD" 
[3] "Experiment1_data_786 n1_786n1_Ca1uM4h_samp_100x 00016125 619.LMD"

Try to pull the latest flowCore from github and let me know if it works for you.

[epgui]

@mikejiang Ohhhhhhhhh my godddd hahaha. Brilliant, thank you so much!

I had been wrestling with this for many hours yesterday and just could not figure it out. I appreciate the quick fix!

[epgui]

I'm getting the same error, but I'm not sure if I've installed the packages correctly. I ran these:

devtools::install_github("RGLab/cytolib") # 2.1.18
devtools::install_github("RGLab/flowCore") # 2.1.2
devtools::install_github("RGLab/flowWorkspace") # 4.1.9

And then again I got:

> gatingset <- flowWorkspace::GatingSet(flowset)
cannot open file '/var/folders/23/z5mf0z_x2sj5xw7m95mdhgyh0000gn/T//Rtmp5xlcs9/filedee7b860abd/data/786 n1/786n1_Ca1uM4h_samp_20x 00016119 613.LMD': No such file or directoryError in h(simpleError(msg, call)) : 
  error in evaluating the argument 'x' in selecting a method for function 'GatingSet': cannot open the connection

[epgui]

Scratch that, I restarted my environment and everything seems to work correctly!