Error in pValues[tmp] <- -pValues[tmp]: NAs are not allowed in subscripted assignments

Dear @varemo,

I currently run into the following error when trying to plot networkPlot or networkPlot2 with class = "distinct" and direction = "both" for a larger set of 5712 genes?

Error in pValues[tmp] <- -pValues[tmp] : 
  NAs are not allowed in subscripted assignments
In addition: Warning messages:
1: In FUN(newX[, i], ...) : no non-missing arguments to min; returning Inf
2: In FUN(newX[, i], ...) : no non-missing arguments to min; returning Inf

Yet, it works fine when I plot networkPlot or networkPlot2 using class = "distinct" but specifying a single direction (direction = "up" or direction = "down").

Strangely, I don't encounter such issues when I run the same code but with a smaller (similar) data set of 968 genes?!

I'm a bit puzzled. Any ideas where the error might be?

Many thanks in advance,

Jan

> sessionInfo()
R version 3.6.1 (2019-07-05)
Platform: x86_64-apple-darwin15.6.0 (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/3.6/Resources/lib/libRlapack.dylib

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

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

other attached packages:
[1] limma_3.40.6        visNetwork_2.0.9    plier_1.54.0        affy_1.62.0         Biobase_2.44.0     
[6] BiocGenerics_0.30.0 piano_2.0.2        

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.4            lattice_0.20-38       relations_0.6-9       gtools_3.8.1          assertthat_0.2.1     
 [6] digest_0.6.25         mime_0.9              slam_0.1-47           R6_2.4.1              ggplot2_3.3.0        
[11] pillar_1.4.3          zlibbioc_1.30.0       gplots_3.0.3          rlang_0.4.5           rstudioapi_0.10      
[16] data.table_1.12.8     gdata_2.18.0          Matrix_1.2-17         DT_0.13               preprocessCore_1.46.0
[21] sets_1.0-18           shinyjs_1.1           BiocParallel_1.18.1   htmlwidgets_1.5.1     igraph_1.2.5         
[26] munsell_0.5.0         shiny_1.4.0.2         fgsea_1.10.1          compiler_3.6.1        httpuv_1.5.2         
[31] pkgconfig_2.0.3       marray_1.62.0         htmltools_0.4.0       tidyselect_0.2.5      tibble_3.0.0         
[36] gridExtra_2.3         fansi_0.4.1           crayon_1.3.4          dplyr_0.8.3           later_1.0.0          
[41] bitops_1.0-6          grid_3.6.1            jsonlite_1.6.1        xtable_1.8-4          gtable_0.3.0         
[46] lifecycle_0.2.0       magrittr_1.5          scales_1.1.0          KernSmooth_2.23-15    cli_2.0.2            
[51] affyio_1.54.0         promises_1.1.0        ellipsis_0.3.0        vctrs_0.2.4           fastmatch_1.1-0      
[56] tools_3.6.1           glue_1.3.2            purrr_0.3.3           yaml_2.2.0            fastmap_1.0.1        
[61] colorspace_1.4-1      BiocManager_1.30.10   cluster_2.1.0         caTools_1.18.0        shinydashboard_0.7.1

Interesting, have not seen this before! I suspect it has to do with some p-values being NA. Could you check your GSAres object and see what $pDistinctDirUp and $pDistinctDirDn are (or $pAdjDistinctDirUp/Dn if using adjusted p-values)? Try setting the ones that are NA to 1. Will this remove the error?

Thanks for your quick response!

I have also already suspected some p-values being NA and checked my input data but there didn't seem to be any NAs in my input data tables.

Regarding my GSAres object, I exported the summary table generated with GSASummaryTable (my_GSAsummaryTable.xlsx). It seems that there are some NAs including $pDistinctDirUp and $pDistinctDirDn for two GO terms.

It also shows up using the following code:

> sum(is.na(gsaRes$pDistinctDirUp))
[1] 2
> sum(is.na(gsaRes$pDistinctDirDn))
[1] 2
> sum(is.na(gsaRes$pAdjDistinctDirUp))
[1] 2
> sum(is.na(gsaRes$pAdjDistinctDirDn))
[1] 2

So I then try to set the NAs to 1 using:

is.na(gsaRes$pDistinctDirUp) <- 1
is.na(gsaRes$pAdjDistinctDirUp) <- 1
is.na(gsaRes$pDistinctDirDn) <- 1
is.na(gsaRes$pAdjDistinctDirDn) <- 1

Unfortunately, I still receive the same error when running:

nw <- networkPlot(gsaRes,
                  class = "distinct",
                  direction = "both",
                  significance=0.01,
                  overlap = 1,
                  lay = 4, # numerical between 1-5 to set the default layout
                  label = "names",
                  ncharLabel=Inf,
                  cexLabel=1.1
)

Strangely, when checking my GSASummaryTable again (after having tried to set NAs to 1 using above code) some additional NA are introduced for the GO term in the first row?!

> sum(is.na(gsaRes$pDistinctDirUp))
[1] 3
> sum(is.na(gsaRes$pDistinctDirDn))
[1] 3
> sum(is.na(gsaRes$pAdjDistinctDirUp))
[1] 3
> sum(is.na(gsaRes$pAdjDistinctDirDn))
[1] 3

I assume that is.na is not properly working on the GSAres object? Any ideas how to get rid of the NAs?

Many thanks for your help! Jan

Just to add to my previous comment. It really seems that is.na(gsaRes$p...) <- 1 introduces an NA at the very first row instead of replacing NA with 1.

I seem to have found a (at least temporary) solution to solve the error.

Replacing NAs in my GSAres object by the following code seems to fix the error when using the networkPlot function with class = "distinct" and direction = "both"

> gsaRes$pDistinctDirUp[is.na(gsaRes$pDistinctDirUp)] <- 1
> sum(is.na(gsaRes$pDistinctDirUp))
[1] 0

Interestingly, networkPlot seems to work fine with only replacing NAs in my gsaRes$pDistinctDirUp by above code without replacing NAs in gsaRes$pDistinctDirDn! In other words, calling the networkPlot function with class = "distinct" and direction = "both" after the following code works fine:

> gsaRes$pDistinctDirUp[is.na(gsaRes$pDistinctDirUp)] <- 1
> sum(is.na(gsaRes$pDistinctDirUp))
[1] 0
> sum(is.na(gsaRes$pDistinctDirDn))
[1] 2

However, interesting is that the error still seems to remain when calling the newer networkPlot2 function instead of networkPlot (even when I replace NA in both, gsaRes$pDistinctDirUp and gsaRes$pDistinctDirDn with 1). Strangely, for the networkPlot2 function with class = "distinct" and direction = "both", the error seems to disappear only when replacing NA with 1 in both, p-values and adjusted p-values, using following code:

> gsaRes$pDistinctDirUp[is.na(gsaRes$pDistinctDirUp)] <- 1
> sum(is.na(gsaRes$pDistinctDirUp))
[1] 0
> gsaRes$pDistinctDirDn[is.na(gsaRes$pDistinctDirDn)] <- 1
> sum(is.na(gsaRes$pDistinctDirDn))
[1] 0
> gsaRes$pAdjDistinctDirUp[is.na(gsaRes$pAdjDistinctDirUp)] <- 1
> sum(is.na(gsaRes$pAdjDistinctDirUp))
[1] 0
> gsaRes$pAdjDistinctDirDn[is.na(gsaRes$pAdjDistinctDirDn)] <- 1
> sum(is.na(gsaRes$pAdjDistinctDirDn))
[1] 0

Any ideas about this behaviour? Many thanks in advance, Jan

varemo / piano

Error in pValues[tmp] <- -pValues[tmp]: NAs are not allowed in subscripted assignments #6