Entering edit mode
I ran into a problem while running the learn.bn() function, and I got the error message candlist[I,1:2] subscript is out of bounds from running the toy example from the pdf guide. I also get this same error message from using the one.step.pigengene function by setting the parameter bnNum=10 as well. I've tried matching the bnlearn package version in the PDF guide and also tried using the latest version (4.9) and I still get this error.
> library(bnlearn)
> learnt <- learn.bn(pigengene=pigengene, bnPath=file.path(saveDir, "bn"),
+ bnNum=10, ## In real applications, at least 100-1000.
+ seed=1, verbose=1)
learn.bn() with bnNum= 10 started at:
2023-10-30 22:35:29.907212
Error in candlist[i, 1:2] : subscript out of bounds
In addition: Warning message:
In discretize(as.data.frame(d), method = "interval", breaks = length(unique(Disease))) :
at least one variable should be continuous
I also have issues with the setting toCompact=TRUE in the one.step.pigengene function. With the parameter set to FALSE, the function runs fine and is able to generate the c5Trees. I was able to run the compact.tree() function standalone.
p1 <- one.step.pigengene(Data=d1,saveDir='pigengene', bnNum=0, verbose=1,
+ seed=1, Labels=Labels, toCompact=TRUE, doHeat=FALSE)
Pigengene started analizing 366 samples using 1000 genes...
[1] "dataNum==1"
Pigengenes...
Pigengene plots in:
\\wsl.localhost/Ubuntu/home/kjngo/pigengene/plots
Making decision trees...
minPerLeaf: 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37
costs:
AML MDS
AML 0 1
MDS 1 0
toCompact: TRUE
Error in get.used.features(c5Tree) :
The class of c5Tree argument must be 'C5.0' !
R session info:
R version 4.3.1 (2023-06-16 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19045)
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
time zone: America/Los_Angeles
tzcode source: internal
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] Pigengene_1.28.0 BiocStyle_2.30.0 graph_1.80.0 BiocGenerics_0.48.0
loaded via a namespace (and not attached):
[1] splines_4.3.1 later_1.3.1 bitops_1.0-7 ggplotify_0.1.2
[5] filelock_1.0.2 tibble_3.2.1 polyclip_1.10-6 preprocessCore_1.64.0
[9] rpart_4.1.21 lifecycle_1.0.3 fastcluster_1.2.3 doParallel_1.0.17
[13] lattice_0.21-9 MASS_7.3-60 backports_1.4.1 magrittr_2.0.3
[17] Hmisc_5.1-1 openxlsx_4.2.5.2 rmarkdown_2.25 yaml_2.3.7
[21] C50_0.1.8 httpuv_1.6.12 zip_2.3.0 cowplot_1.1.1
[25] DBI_1.1.3 RColorBrewer_1.1-3 zlibbioc_1.48.0 purrr_1.0.2
[29] ggraph_2.1.0 RCurl_1.98-1.12 nnet_7.3-19 yulab.utils_0.1.0
[33] tweenr_2.0.2 rappdirs_0.3.3 GenomeInfoDbData_1.2.11 IRanges_2.36.0
[37] S4Vectors_0.40.1 enrichplot_1.22.0 ggrepel_0.9.4 tidytree_0.4.5
[41] gdata_3.0.0 reactome.db_1.86.0 pheatmap_1.0.12 codetools_0.2-19
[45] DOSE_3.28.0 ggforce_0.4.1 tidyselect_1.2.0 aplot_0.2.2
[49] farver_2.1.1 viridis_0.6.4 base64enc_0.1-3 dynamicTreeCut_1.63-1
[53] matrixStats_1.0.0 stats4_4.3.1 BiocFileCache_2.10.1 jsonlite_1.8.7
[57] ellipsis_0.3.2 tidygraph_1.2.3 Formula_1.2-5 iterators_1.0.14
[61] survival_3.5-7 foreach_1.5.2 tools_4.3.1 treeio_1.26.0
[65] HPO.db_0.99.2 Rcpp_1.0.11 glue_1.6.2 gridExtra_2.3
[69] xfun_0.40 qvalue_2.34.0 bnlearn_4.8.1 GenomeInfoDb_1.38.0
[73] dplyr_1.1.3 withr_2.5.2 BiocManager_1.30.22 fastmap_1.1.1
[77] fansi_1.0.5 digest_0.6.33 R6_2.5.1 mime_0.12
[81] gridGraphics_0.5-1 colorspace_2.1-0 GO.db_3.18.0 gtools_3.9.4
[85] RSQLite_2.3.2 inum_1.0-5 utf8_1.2.4 tidyr_1.3.0
[89] generics_0.1.3 data.table_1.14.8 htmlwidgets_1.6.2 graphlayouts_1.0.1
[93] httr_1.4.7 scatterpie_0.2.1 graphite_1.48.0 pkgconfig_2.0.3
[97] gtable_0.3.4 blob_1.2.4 impute_1.76.0 XVector_0.42.0
[101] clusterProfiler_4.10.0 shadowtext_0.1.2 htmltools_0.5.6.1 fgsea_1.28.0
[105] scales_1.2.1 Biobase_2.62.0 png_0.1-8 ggfun_0.1.3
[109] knitr_1.45 rstudioapi_0.15.0 reshape2_1.4.4 checkmate_2.3.0
[113] nlme_3.1-163 curl_5.1.0 cachem_1.0.8 stringr_1.5.0
[117] BiocVersion_3.18.0 parallel_4.3.1 HDO.db_0.99.1 libcoin_1.0-10
[121] foreign_0.8-85 AnnotationDbi_1.64.0 ReactomePA_1.46.0 pillar_1.9.0
[125] grid_4.3.1 vctrs_0.6.4 promises_1.2.1 dbplyr_2.4.0
[129] cluster_2.1.4 xtable_1.8-4 htmlTable_2.4.2 Rgraphviz_2.46.0
[133] evaluate_0.22 Cubist_0.4.2.1 mvtnorm_1.2-3 cli_3.6.1
[137] compiler_4.3.1 rlang_1.1.1 crayon_1.5.2 plyr_1.8.9
[141] fs_1.6.3 stringi_1.7.12 WGCNA_1.72-1 viridisLite_0.4.2
[145] BiocParallel_1.36.0 MPO.db_0.99.7 munsell_0.5.0 Biostrings_2.70.1
[149] lazyeval_0.2.2 GOSemSim_2.28.0 Matrix_1.6-1.1 patchwork_1.1.3
[153] bit64_4.0.5 ggplot2_3.4.4 KEGGREST_1.42.0 shiny_1.7.5.1
[157] interactiveDisplayBase_1.40.0 AnnotationHub_3.10.0 partykit_1.2-20 igraph_1.5.1
[161] memoise_2.0.1 ggtree_3.10.0 fastmatch_1.1-4 bit_4.0.5
[165] ape_5.7-1 gson_0.1.0
Thank you for your time and help!
Thank you Dr. Zare for getting back to my inquires.
Below is my full output I also ran the learn.bn() command with higher detail (verbose =4). The pigengene object should be the same as you had shown above. I'm wondering if there is an issue with paths since I am running this on a windows machine?
Your guess is true. We have not test
learn.bnon Windows. Are you running it on a virtual Ubuntu on Windows? The path shown in the line above the error message is weird. The following can help me to diagnose as I do not have access to a Windows machine:Also, can you rerun with
verbos=6?Thank you Dr. Zare. I have WSL2 (Windows Subsystem for Linux) architecture installed. I ran the commands as you mentioned below are the screenshots of the outputs.
Interestingly, I also tried to change the path of the working directory where the code for the demo was being executed and it seems the learn.bn() function was able to work this time around. I am not sure what changed, as previously, this step was not working even with a different working directory. The only change I noticed that was different from my session info was that the bnlearn package was listed under "other attached packages" instead of the "loaded via a namespace (and not attached)" like above. Not sure if this makes a difference, and if it does how do I ensure the bnlearn package is attached when executing the code? Thank you for your time and help.
The output of
print(file.path(saveDir, "test"))is so unexpected that I do not know how to proceed with this mix of Windows and Linux. My suggestion is that you use a real Linux system and if it is not available, use a pure Windows system. Can you do this? If not, I am still willing to help, but that would require much more work, e.g., we can start with rerunning the example withverbose=10using Pigengene 1.29.10, or later.The bnlearn package came from "Loaded via a namespace (and not attached)" to "other attached packages" because you probably called it using
library(bnlearn)somewhere in your newer R session, but not in the other one.I think your second attempt worked because you changed your working directory to a normal Windows path (i.e.,
C:\\Users\\...) rather than the weird\\\\wsl.localhost\\Ubuntu...combo path. Also, there might have been some permission issues in that path.Thanks Dr. Zare. I will run Pigengene 1.29.10 or later in either an pure linux or pure windows environment.