Loss of ability to use gates in imported workspace without transforming the data

[PedroMilanezAlmeida]

A few days ago I updated to R 4.0.2 and Bioc 3.11 and lost the ability to parse a FlowJo workspace using flowjo_to_gatingset using transform = FALSE while also applying the gates.

The feature described above was particularly useful when dealing with data that were transformed outside FlowJo but imported into FlowJo for gating. For us, this functionality is very important for analysis of CITEseq data.

I wonder whether it would be possible to re-enable the use of gates without data transformation (i.e., get "raw gates").

Here are the gates in FlowJo and in R:

In case helpful, for the gate above, the info in the wsp/xml file is:

This is the info from R:

[PedroMilanezAlmeida]

> sessionInfo()
R version 4.0.2 (2020-06-22)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Mojave 10.14.6

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.0/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

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

other attached packages:
 [1] ggcyto_1.16.0             ncdfFlow_2.34.0           BH_1.72.0-3              
 [4] RcppArmadillo_0.9.900.2.0 ggplot2_3.3.2             flowCore_2.0.1           
 [7] flowUtils_1.52.0          CytoML_2.0.5              flowWorkspace_4.0.6      
[10] FlowSOM_1.20.0            igraph_1.2.5              XML_3.99-0.5             
[13] magrittr_1.5              glue_1.4.1                Rcpp_1.0.5               
[16] gridExtra_2.3             pheatmap_1.0.12           foreach_1.5.0            
[19] BiocManager_1.30.10      

loaded via a namespace (and not attached):
 [1] lattice_0.20-41             corpcor_1.6.9               RProtoBufLib_2.0.0         
 [4] png_0.1-7                   digest_0.6.25               R6_2.4.1                   
 [7] plyr_1.8.6                  stats4_4.0.2                pillar_1.4.6               
[10] zlibbioc_1.34.0             rlang_0.4.7                 rstudioapi_0.11            
[13] data.table_1.13.0           Rgraphviz_2.32.0            hexbin_1.28.1              
[16] RUnit_0.4.32                labeling_0.3                stringr_1.4.0              
[19] munsell_0.5.0               compiler_4.0.2              pkgconfig_2.0.3            
[22] ConsensusClusterPlus_1.52.0 BiocGenerics_0.34.0         base64enc_0.1-3            
[25] tidyselect_1.1.0            tibble_3.0.3                codetools_0.2-16           
[28] matrixStats_0.56.0          crayon_1.3.4                dplyr_1.0.1                
[31] withr_2.2.0                 grid_4.0.2                  RBGL_1.64.0                
[34] tsne_0.1-3                  jsonlite_1.7.0              gtable_0.3.0               
[37] lifecycle_0.2.0             scales_1.1.1                graph_1.66.0               
[40] RcppParallel_5.0.2          stringi_1.4.6               farver_2.0.3               
[43] latticeExtra_0.6-29         xml2_1.3.2                  ellipsis_0.3.1             
[46] generics_0.0.2              vctrs_0.3.2                 RColorBrewer_1.1-2         
[49] iterators_1.0.12            tools_4.0.2                 Biobase_2.48.0             
[52] purrr_0.3.4                 jpeg_0.1-8.1                parallel_4.0.2             
[55] yaml_2.2.1                  colorspace_1.4-1            cluster_2.1.0              
[58] cytolib_2.0.3

[PedroMilanezAlmeida]

What is weird is that, on a different dataset (simulated in R, imported into FJ but no FJ-data transformation applied [linear]), flowjo_to_gatingset does NOT distort the gate coordinates:

And, in the first dataset (CITEseq) the first gate is NOT distorted neither (CD11b-CD14-, not shown here).

[PedroMilanezAlmeida]

@gfinak Thanks for your reply!

[mikejiang]

transform = FALSE is meant to only parse the raw gate(for troubleshooting purpose) without doing any gating or data loading.
It is only valid when execute = FALSE, which is not relevant to what you are trying to do. In fact, you should see a message when you try to disable transform while still attempt to proceed parsing

> gs <- flowjo_to_gatingset(ws, name = 4, transform = F, subset = 1)
'transform = false' is ignored when 'is_gating' is set to true!
> gs
A GatingSet with 1 samples

You shouldn't need to fiddle with transform flag, the parser is supposed to import whatever is recorded in flowJo workspace, doesn't matter whether data and gates are transformed or linear.

The right question is why your first gate is distorted, which may require a reproducible example to trouble shoot. (feel free to zip it and send it to [email protected])

[PedroMilanezAlmeida]

Hey Mike, thanks for your comment.

I understand that transform was only meant to be used in troubleshooting. However, until now, often when I had non-standard/non-FCS data that I needed to analyze in FlowJo and reimport in R, turning off transform was the only way for the gates to be parsed correctly.

Thanks again for your help, I will send you a reproducible example via email later today.

[mikejiang]

Again, It has nothing to do with transformation since all the channels are on linear scale. It turns out to be the gate extension issue, which is discussed in #106 . Before we figure out the more robust way to automatically determine whether to trigger the gate extension (to the far lower end of axis), you can manually turn it off (by lowering the threshold extend_val) to get your correct gates back.

gs <- flowjo_to_gatingset(ws,
                                  name = 1,
                                  subset = "CITEseq_downsampled.fcs",
                                  sampNloc = "sampleNode"
                          , extend_val = -Inf
      )
autoplot(gs[[1]],
         gs_get_pop_paths(gs[[1]])[4])

> gh_pop_compare_stats(gs[[1]])
   openCyto.freq xml.freq openCyto.count xml.count        node
1:     1.0000000 1.000000           5000      5000        root
2:     1.0000000 1.000000           5000      5000   all cells
3:     0.8800000 0.882200           4400      4411 CD11b-CD14-
4:     0.1488636 0.148039            655       653   CD19-CD3-

[PedroMilanezAlmeida]

Cool!!! Thank you so much! I'll give it try and let you know how it goes! Thanks again!

EDIT: @mikejiang it solved my issue!