Currently, I am attempting to compensate for spillover across channels in mass cytometry datasets. However, whenever I attempt to do so with files that were manually cleaned to remove debris and aggregates, the spillover matrix that is generated appears to have values greater than 1. I have absolutely no idea why this would be the case even after looking at the source code and documentation for the computeSpillmat function and CATALYST, and consequently I would appreciate any and all assistance to resolving this issue.

library(CATALYST)
library(cowplot)
library(flowCore)
library(ggplot2)
library(SingleCellExperiment)
library(readxl)
library(FlowSOM)

listfcs=c("20220318_Yukako_Sample1_Cleaned.fcs", "20220325_Stimulation_Cleaned.fcs","March_30_Cleaned_sample_1.fcs","March_30_Cleaned_sample_2.fcs", "March_30_Cleaned_sample_3.fcs","March_30_Cleaned_sample_4.fcs",
          "April_5_Cleaned_sample_1.fcs","April_5_Cleaned_sample_2.fcs", "April_5_Cleaned_sample_3.fcs","April_5_Cleaned_sample_4.fcs",
          "April_11_Cleaned_sample_1.fcs","April_11_Cleaned_sample_2.fcs", "April_11_Cleaned_sample_3.fcs","April_11_Cleaned_sample_4.fcs")
realdata <- read.flowSet(files=listfcs, path="C:/Users/DWQ/OneDrive - Mass General Brigham/Documents/Raw data/Cleaned")
setwd("C:/Users/DWQ/OneDrive - Mass General Brigham/Documents")
panel <- "Panel_markers.xlsx"
barkey <- "Fluidigm_20plex_barcode_key.csv"
panel <- read_excel(panel)
head(data.frame(panel))

md <- "Cleaned_metadata.xlsx"

md <- read_excel(md)
md$condition <- factor(md$condition, levels = c("Ref", "Radiotherapy"))
md$sample_id <- factor(md$sample_id, levels = md$sample_id[order(md$condition)])
set.seed(1234)
sce <- prepData(realdata, panel, md, features = panel$fcs_colname)
bc_ms <- c(89,103,120, 127,131,133,138, 140:156, 158:176,191,193:195,198,208,209)

sce <- assignPrelim(sce, bc_ms, verbose = FALSE)
sce <- applyCutoffs(estCutoffs(sce))

# compute & extract spillover matrix
sce <- computeSpillmat(sce, assay="exprs")
sm <- metadata(sce)$spillover_matrix

sce <- compCytof(sce, sm, method = "nnls", overwrite = TRUE)

sce <- cluster(sce, features = "state",
               xdim = 10, ydim = 10, maxK = 20, seed = 1234)
sce <- runDR(sce, "UMAP", cells = 1e3, features = "state")
sce <- runDR(sce, "TSNE", cells = 500, features = "state")

This is the resulting error:

> sce <- applyCutoffs(estCutoffs(sce))
Error in optim(startVec, opfct, hessian = TRUE, method = optMethod, control = list(maxit = maxIt,  : 
  non-finite finite-difference value [3]
Error in optim(startVec, opfct, hessian = TRUE, method = optMethod, control = list(maxit = maxIt,  : 
  non-finite finite-difference value [3]
Error in optim(startVec, opfct, hessian = TRUE, method = optMethod, control = list(maxit = maxIt,  : 
  non-finite finite-difference value [3]
Error in optim(startVec, opfct, hessian = TRUE, method = optMethod, control = list(maxit = maxIt,  : 
  non-finite finite-difference value [3]

> # compute & extract spillover matrix
> sce <- computeSpillmat(sce, assay="exprs")

> sm <- metadata(sce)$spillover_matrix

> sce <- compCytof(sce, sm, method = "nnls", overwrite = TRUE)
Error in .check_sm(x, isotope_list) : 
The supplied spillover matrix is invalid as it contains entries greater than 1.
Valid spill values are non-negative and mustn't exceed 1.

This is a list of the packages I am using:

sessionInfo()
R version 4.2.1 (2022-06-23 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19042)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.utf8  LC_CTYPE=English_United States.utf8    LC_MONETARY=English_United States.utf8
[4] LC_NUMERIC=C                           LC_TIME=English_United States.utf8    

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

other attached packages:
 [1] FlowSOM_2.4.0               igraph_1.3.4                readxl_1.4.0                ggplot2_3.3.6              
 [5] flowCore_2.8.0              cowplot_1.1.1               CATALYST_1.20.1             SingleCellExperiment_1.18.0
 [9] SummarizedExperiment_1.26.1 Biobase_2.56.0              GenomicRanges_1.48.0        GenomeInfoDb_1.32.2        
[13] IRanges_2.30.0              S4Vectors_0.34.0            BiocGenerics_0.42.0         MatrixGenerics_1.8.1       
[17] matrixStats_0.62.0         

loaded via a namespace (and not attached):
  [1] backports_1.4.1             circlize_0.4.15             drc_3.0-1                   plyr_1.8.7                 
  [5] ConsensusClusterPlus_1.60.0 splines_4.2.1               BiocParallel_1.30.3         scattermore_0.8            
  [9] scater_1.24.0               TH.data_1.1-1               digest_0.6.29               foreach_1.5.2              
 [13] viridis_0.6.2               fansi_1.0.3                 magrittr_2.0.3              ScaledMatrix_1.4.0         
 [17] CytoML_2.8.0                cluster_2.1.3               doParallel_1.0.17           aws.signature_0.6.0        
 [21] ComplexHeatmap_2.12.0       RcppParallel_5.1.5          sandwich_3.0-2              flowWorkspace_4.8.0        
 [25] cytolib_2.8.0               jpeg_0.1-9                  colorspace_2.0-3            ggrepel_0.9.1              
 [29] dplyr_1.0.9                 crayon_1.5.1                RCurl_1.98-1.7              jsonlite_1.8.0             
 [33] hexbin_1.28.2               graph_1.74.0                survival_3.3-1              zoo_1.8-10                 
 [37] iterators_1.0.14            glue_1.6.2                  polyclip_1.10-0             gtable_0.3.0               
 [41] nnls_1.4                    zlibbioc_1.42.0             XVector_0.36.0              GetoptLong_1.0.5           
 [45] DelayedArray_0.22.0         ggcyto_1.24.1               car_3.1-0                   BiocSingular_1.12.0        
 [49] Rgraphviz_2.40.0            shape_1.4.6                 abind_1.4-5                 scales_1.2.0               
 [53] pheatmap_1.0.12             mvtnorm_1.1-3               DBI_1.1.3                   rstatix_0.7.0              
 [57] Rcpp_1.0.9                  plotrix_3.8-2               viridisLite_0.4.0           clue_0.3-61                
 [61] rsvd_1.0.5                  httr_1.4.3                  RColorBrewer_1.1-3          ellipsis_0.3.2             
 [65] pkgconfig_2.0.3             XML_3.99-0.10               farver_2.1.1                scuttle_1.6.2              
 [69] deldir_1.0-6                utf8_1.2.2                  tidyselect_1.1.2            rlang_1.0.4                
 [73] reshape2_1.4.4              cellranger_1.1.0            munsell_0.5.0               tools_4.2.1                
 [77] cli_3.3.0                   generics_0.1.3              broom_1.0.0                 ggridges_0.5.3             
 [81] aws.s3_0.3.21               stringr_1.4.0               yaml_2.3.5                  purrr_0.3.4                
 [85] RBGL_1.72.0                 sparseMatrixStats_1.8.0     xml2_1.3.3                  compiler_4.2.1             
 [89] rstudioapi_0.13             beeswarm_0.4.0              curl_4.3.2                  png_0.1-7                  
 [93] ggsignif_0.6.3              tibble_3.1.8                tweenr_1.0.2                stringi_1.7.8              
 [97] lattice_0.20-45             Matrix_1.4-1                vctrs_0.4.1                 pillar_1.8.0               
[101] lifecycle_1.0.1             GlobalOptions_0.1.2         BiocNeighbors_1.14.0        data.table_1.14.2          
[105] bitops_1.0-7                irlba_2.3.5                 colorRamps_2.3.1            R6_2.5.1                   
[109] latticeExtra_0.6-30         gridExtra_2.3               RProtoBufLib_2.8.0          vipor_0.4.5                
[113] codetools_0.2-18            MASS_7.3-58                 gtools_3.9.3                assertthat_0.2.1           
[117] rjson_0.2.21                withr_2.5.0                 multcomp_1.4-19             GenomeInfoDbData_1.2.8     
[121] parallel_4.2.1              ncdfFlow_2.42.1             grid_4.2.1                  beachmat_2.12.0            
[125] tidyr_1.2.0                 ggpointdensity_0.1.0        DelayedMatrixStats_1.18.0   carData_3.0-5              
[129] Rtsne_0.16                  ggpubr_0.4.0                ggnewscale_0.4.7            ggforce_0.3.3              
[133] base64enc_0.1-3             ggbeeswarm_0.6.0            interp_1.1-3

Please let me know if you need any additional information.

Hi,

There is a first error at sce <- applyCutoffs(estCutoffs(sce)). It should be solved before considering the next error, ie the spillover matrix.

In fact, the first error occurs because assignPrelim() was not given a valid debarcoding scheme. The debarcoding scheme describes the masses used for barcoding the samples, not the ones used for characterizing the cells. From CATALYST documentation, the debarcoding scheme is a binary table with sample IDs as row and numeric barcode masses as column names:

   102 104 105 106 108 110
A1   1   1   1   0   0   0
A2   1   1   0   1   0   0
A3   1   1   0   0   1   0
A4   1   1   0   0   0   1
A5   1   0   1   1   0   0
B1   1   0   1   0   1   0

This is typically created using Excel or so. There is an example of a barcoded FCS and a debarcoding CSV scheme at Nolan's lab as pointed by PF Gherardini debarcoding tool in Premessa.

Once this is solved, I guess the errors will not persist. If the data come from a core facility nearby, ask them.

Best, Samuel