error in extractSignatures

[LuyaoRen]

Dear Poison Allen,

when I run extractSignatures, I got some errors and didn't know how to solve it. Can you help me?

Error in (function (...) : All the runs produced an error:
-#1 [r=2] -> NMF::nmf - 10/10 fit(s) threw an error.

Error(s) thrown:

• run writeup/DOI #1: non-conformable arrays
  -VCF to MAF conversion? #2 [r=3] -> NMF::nmf - 10/10 fit(s) threw an error.

Error(s) thrown:

• run writeup/DOI #1: non-conformable arrays
  -feature request: rainfall plot #3 [r=4] -> NMF::nmf - 10/10 fit(s) threw an error.

Error(s) thrown:

• run writeup/DOI #1: non-conformable arrays
  -Dockerizeit #4 [r=5] -> NMF::nmf - 10/10 fit(s) threw an error.

Error(s) thrown:

• run writeup/DOI #1: non-conformable arrays
  -CNV/RNAseq data into oncoplot #5 [r=6] -> NMF::nmf - 10/10 fit(s) threw an error.

Error(s) thrown:

• run writeup/DOI #1: non-conformable arrays

[PoisonAlien]

Hi,

How many samples do you have ? Can post your commands ?

[OmichessanH]

Hi;
I got the same error; with the example provided in the bioconductor vignette using TCGA laml.
laml.sign = extractSignatures(mat = laml.tnm, nTry = 6, plotBestFitRes = FALSE)

What is the issue?

[PoisonAlien]

Oh that’s weird ! Try loading nmf library manually and rerun the example. Also what os are you on !? Can you post your sessioninfo!?

[OmichessanH]

Thanks for your quick answer !!
Loading manually NMF give the same error

My sessionInfo() :

R version 3.4.3 (2017-11-30)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS: /usr/local/R-3.4.3/lib64/R/lib/libRblas.so
LAPACK: /usr/local/R-3.4.3/lib64/R/lib/libRlapack.so

locale:
[1] LC_CTYPE=fr_FR.UTF-8 LC_NUMERIC=C
[3] LC_TIME=fr_FR.UTF-8 LC_COLLATE=fr_FR.UTF-8
[5] LC_MONETARY=fr_FR.UTF-8 LC_MESSAGES=fr_FR.UTF-8
[7] LC_PAPER=fr_FR.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=fr_FR.UTF-8 LC_IDENTIFICATION=C

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

other attached packages:
[1] doParallel_1.0.11 iterators_1.0.8 foreach_1.4.3
[4] maftools_1.4.05 NMF_0.20.6 Biobase_2.38.0
[7] BiocGenerics_0.24.0 cluster_2.0.6 rngtools_1.2.4
[10] pkgmaker_0.22 registry_0.5

loaded via a namespace (and not attached):
[1] RMySQL_0.10.13 bit64_0.9-7
[3] splines_3.4.3 assertthat_0.2.0
[5] stats4_3.4.3 blob_1.1.0
[7] BSgenome_1.46.0 GenomeInfoDbData_0.99.1
[9] Rsamtools_1.30.0 slam_0.1-40
[11] progress_1.1.2 ggrepel_0.7.0
[13] RSQLite_2.0 lattice_0.20-35
[15] digest_0.6.12 GenomicRanges_1.30.0
[17] RColorBrewer_1.1-2 XVector_0.18.0
[19] colorspace_1.3-2 cowplot_0.9.1
[21] Matrix_1.2-12 plyr_1.8.4
[23] XML_3.98-1.9 GetoptLong_0.1.6
[25] biomaRt_2.34.0 zlibbioc_1.24.0
[27] xtable_1.8-2 scales_0.5.0
[29] BiocParallel_1.12.0 tibble_1.3.4
[31] IRanges_2.12.0 ggplot2_2.2.1
[33] SummarizedExperiment_1.8.0 GenomicFeatures_1.30.0
[35] lazyeval_0.2.1 mclust_5.4
[37] survival_2.41-3 magrittr_1.5
[39] memoise_1.1.0 changepoint_2.2.2
[41] tools_3.4.3 data.table_1.10.4-3
[43] prettyunits_1.0.2 GlobalOptions_0.0.12
[45] matrixStats_0.52.2 gridBase_0.4-7
[47] ComplexHeatmap_1.17.1 stringr_1.2.0
[49] S4Vectors_0.16.0 munsell_0.4.3
[51] DelayedArray_0.4.1 AnnotationDbi_1.40.0
[53] Biostrings_2.46.0 compiler_3.4.3
[55] GenomeInfoDb_1.14.0 rlang_0.1.4
[57] grid_3.4.3 RCurl_1.95-4.8
[59] VariantAnnotation_1.24.2 rjson_0.2.15
[61] circlize_0.4.2 bitops_1.0-6
[63] gtable_0.2.0 codetools_0.2-15
[65] DBI_0.7 reshape2_1.4.2
[67] R6_2.2.2 gridExtra_2.3
[69] zoo_1.8-0 GenomicAlignments_1.14.1
[71] rtracklayer_1.38.1 bit_1.1-12
[73] shape_1.4.3 stringi_1.1.6
[75] Rcpp_0.12.14 wordcloud_2.5

[PoisonAlien]

Okay this looks similar to the issue posted on NMF github page renozao/NMF#89

I guess your input data has NAs (?) which should't be. Can you attach your matrix here as a text file so that it helps me to see what exactly is the issue?

write.table(x = laml.tnm$nmf_matrix, file = "tnm.tsv", sep = "\t", quote = FALSE)
#Attach tnm.tsv

[OmichessanH]

tnm.txt

There is no NAs in the matrix. Please find enclosed the matrix. Thanks for your help

[PoisonAlien]

Hi,
It runs correctly on my system without any errors. But I see that nmf package version on your system is 0.20.6 (which is also on CRAN). Try installing NMF from devel branch of GitHub page and see if it helps.

devtools::install_github(repo = "renozao/NMF", ref = "devel")

Please let me know if this works.

[OmichessanH]

Hi;
It finally works with NMF from devel branch.

Thank you

[clersdom]

Hi, I have the same error and can not see any NAs or rows with all entries 0, but I do see a column with all values 0... I think that could be the error?
I have also tried installing NMF again from Github and I am still getting the same error.
It works for a bigger subset of samples (n=85), but not for the current set with 33 samples for which I am having the error.

my.sign <-extractSignatures(mat = my.tnm, n=3)

-Found zero mutations for conversions:
A[T>G]A
-Running NMF for factorization rank: 3
Error: NMF::nmf - Input matrix x contains at least one null or NA-filled row.

Indeed when I look at the nmf matrix I can see that the "A[T>G]A" column has 0s for all the rows. Is there a way to fix this? Could I manually add a smaller value like 0.001 in that column to perhaps try skip this message? Not to sure how to do so within a list object my.tnm$nmf_matrix

I have run this same commands a year ago on the same samples and data and it worked fine! So I would like to reproduce the results again...

sessionInfo()
R version 3.6.1 (2019-07-05)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS High Sierra 10.13.6

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_GB.UTF-8/en_GB.UTF-8/en_GB.UTF-8/C/en_GB.UTF-8/en_GB.UTF-8

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

other attached packages:
[1] doParallel_1.0.15 iterators_1.0.12 foreach_1.5.0 ggpubr_0.3.0
[5] forcats_0.5.0 stringr_1.4.0 dplyr_1.0.0 purrr_0.3.4
[9] readr_1.3.1 tidyr_1.1.0 tibble_3.0.1 ggplot2_3.3.1
[13] tidyverse_1.3.0 NMF_0.23.6 Biobase_2.46.0 cluster_2.1.0
[17] rngtools_1.5 pkgmaker_0.31 registry_0.5-1 BSgenome.Hsapiens.UCSC.hg19_1.4.0
[21] BSgenome_1.54.0 rtracklayer_1.46.0 Biostrings_2.54.0 XVector_0.26.0
[25] GenomicRanges_1.38.0 GenomeInfoDb_1.22.1 IRanges_2.20.2 S4Vectors_0.24.4
[29] BiocGenerics_0.32.0 maftools_2.2.10

loaded via a namespace (and not attached):
[1] colorspace_1.4-1 ggsignif_0.6.0 ellipsis_0.3.1 rio_0.5.16
[5] rprojroot_1.3-2 fs_1.4.1 rstudioapi_0.11 remotes_2.1.1
[9] bit64_0.9-7 AnnotationDbi_1.48.0 fansi_0.4.1 lubridate_1.7.9
[13] xml2_1.3.2 codetools_0.2-16 splines_3.6.1 knitr_1.28
[17] pkgload_1.1.0 jsonlite_1.6.1 Rsamtools_2.2.3 broom_0.5.6
[21] gridBase_0.4-7 dbplyr_1.4.4 BiocManager_1.30.10 compiler_3.6.1
[25] httr_1.4.1 backports_1.1.7 assertthat_0.2.1 Matrix_1.2-18
[29] cli_2.0.2 prettyunits_1.1.1 tools_3.6.1 gtable_0.3.0
[33] glue_1.4.1 GenomeInfoDbData_1.2.2 reshape2_1.4.4 Rcpp_1.0.4.6
[37] carData_3.0-4 cellranger_1.1.0 vctrs_0.3.1 nlme_3.1-148
[41] xfun_0.14 ps_1.3.3 testthat_2.3.2 openxlsx_4.1.5
[45] rvest_0.3.5 lifecycle_0.2.0 devtools_2.3.0 rstatix_0.5.0
[49] XML_3.99-0.3 zlibbioc_1.32.0 scales_1.1.1 hms_0.5.3
[53] SummarizedExperiment_1.16.1 RColorBrewer_1.1-2 curl_4.3 memoise_1.1.0
[57] RSQLite_2.2.0 stringi_1.4.6 desc_1.2.0 pkgbuild_1.0.8
[61] zip_2.0.4 BiocParallel_1.20.1 bibtex_0.4.2.2 rlang_0.4.6
[65] pkgconfig_2.0.3 matrixStats_0.56.0 bitops_1.0-6 lattice_0.20-41
[69] GenomicAlignments_1.22.1 bit_1.1-15.2 processx_3.4.2 tidyselect_1.1.0
[73] plyr_1.8.6 magrittr_1.5 R6_2.4.1 generics_0.0.2
[77] DelayedArray_0.12.3 DBI_1.1.0 pillar_1.4.4 haven_2.3.1
[81] foreign_0.8-76 withr_2.2.0 survival_3.1-12 abind_1.4-5
[85] RCurl_1.98-1.2 modelr_0.1.8 crayon_1.3.4 car_3.0-8
[89] wordcloud_2.6 usethis_1.6.1 grid_3.6.1 readxl_1.3.1
[93] data.table_1.12.8 callr_3.4.3 blob_1.2.1 reprex_0.3.0
[97] digest_0.6.25 xtable_1.8-4 munsell_0.5.0 sessioninfo_1.1.1

[PoisonAlien]

Hi, There is an argument pConstant which you can use to add small positive value to avoid zeros. Set it to your desired value e.g, pConstant=0.01.
Let me know if you still have the issue.

[clersdom]

Hi, that did the trick! Also for the estimateSignatures function. Should it not be better to add 0.001? Does it make any difference at all?

Thanks for your fast reply.

[PoisonAlien]

I have not really observed significant changes between 0.01 or 0.001. Sometimes people even use 1e-12. I think it should be fine in anyways.