Open friedue opened 5 years ago
Hey Friederike! Thanks for trying out dyno. This is a known bug that I should really look into urgently. The first time you download a TI method, you get a error message. The solution is to simply try again :)
Could you try running it again and see if it works? If it does, don't close this ticket yet, that'll remind me to take a look at this problem.
Robrecht
tried it again:
Error traceback:
1:
Attaching package: ‘dplyr’
The following objects are masked from ‘package:stats’:
filter, lag
The following objects are masked from ‘package:base’:
intersect, setdiff, setequal, union
Attaching package: ‘purrr’
The following object is masked from ‘package:jsonlite’:
flatten
Warning messages:
1: In rgl.init(initValue, onlyNULL) : RGL: unable to open X11 display
2: 'rgl_init' failed, running with rgl.useNULL = TRUE
Killed
Error: $ operator is invalid for atomic vectors
EDIT: realized I had forgotten to actually include the error...
Are you still at the conference? Do you have time at 8 to show me your problem?
Sure!
just for reference:
dim(counts)
[1] 15520 7263
and more reference:
model <- infer_trajectory(dataset,"projected_paga")
Error traceback:
1: Killed
Error: $ operator is invalid for atomic vectors
In regard to the problem that may occur because of the normalization step that slingshot suggests -- maybe you could use the slot from your dataset file that contains the expression as is assuming that the submitter has normalized those values the way they want them to be normalized?
Hi There! I am getting the same problem with the
Error: $ operator is invalid for atomic vectors
Is there a solution for this? It has happened with every TI method I have tried using dyno, and with slingshot it is also accompanied by the same error in this thread:
Warning messages: 1: In rgl.init(initValue, onlyNULL) : RGL: unable to open X11 display 2: 'rgl_init' failed, running with rgl.useNULL = TRUE Killed
Happy to show errors etc but its the same thing - any solution suggestions?
Hi Friederike,
The new version of dyno should be able to handle sparse matrices much better. Only the software of the individual TI methods might not support sparse matrices. Could you install the latest version of dyno and try again?
Robrecht
@abhaduri You are likely having a different issue. There was a time where dyno was not forwarding errors generated within containers correctly, but this has since been fixed. Could you reinstall and try again? If the problem persists, feel free to create a new issue.
congrats on the publication!
I've re-installed dyno, this is what's happening now:
> library(dyno)
> dataset <- wrap_expression(counts = cts,
expression = logcounts(sf))
> model <- infer_trajectory(dataset,"slingshot", verbose = TRUE)
Executing 'slingshot' on '20190403_232308__data_wrapper__2g1fBJWQ6V'
With parameters: list(shrink = 1L, reweight = TRUE, reassign = TRUE, thresh = 0.001, maxit = 10L, stretch = 2L, smoother = "smooth.spline", shrink.method = "cosine"),
inputs: counts, and
priors :
Input saved to /tmp/folders/xw/962yzh952d1_jhtnnf435j5r0000h2/T//RtmpyxWO9m/filea97d72393a51/ti
Running /usr/local/bin/docker run -e 'TMPDIR=/tmp2' --workdir /ti/workspace -v '/tmp/folders/xw/962yzh952d1_jhtnnf435j5r0000h2/T//RtmpyxWO9m/filea97d72393a51/ti:/ti' -v \
'/tmp/folders/xw/962yzh952d1_jhtnnf435j5r0000h2/T//RtmpyxWO9m/filea97db23d66d/tmp:/tmp2' 'dynverse/ti_slingshot:v0.9.9' --dataset /ti/input.h5 --output /ti/output.h5
Loading required package: dynutils
Error: Error during trajectory inference
Loading required package: dynutils
> str(logcounts(sf))
num [1:17281, 1:13011] 0.8 0.8 1.31 0.8 0.8 ...
- attr(*, "dimnames")=List of 2
..$ : chr [1:17281] "ENSMUSG00000033845" "ENSMUSG00000025903" "ENSMUSG00000033793" "ENSMUSG00000025907" ...
..$ : chr [1:13011] "DB1" "DB2" "DB3" "DB4" ...
> str(cts)
num [1:17281, 1:13011] 1 1 2 1 1 1 3 17 1 3 ...
- attr(*, "dimnames")=List of 2
..$ : chr [1:17281] "ENSMUSG00000033845" "ENSMUSG00000025903" "ENSMUSG00000033793" "ENSMUSG00000025907" ...
..$ : chr [1:13011] "DB1" "DB2" "DB3" "DB4" ...
> sessionInfo()
R version 3.5.1 (2018-07-02)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS 10.14
Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.5/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] parallel stats4 stats graphics grDevices utils datasets methods base
other attached packages:
[1] dyno_0.9.9 dynwrap_1.0.0 dynplot_1.0.0 dynmethods_1.0.0 dynguidelines_1.0.0 dynfeature_1.0.0 magrittr_1.5
[8] scater_1.9.23 ggplot2_3.1.0 SingleCellExperiment_1.3.11 SummarizedExperiment_1.10.1 DelayedArray_0.6.6 BiocParallel_1.14.2 matrixStats_0.54.0
[15] Biobase_2.40.0 GenomicRanges_1.32.6 GenomeInfoDb_1.16.0 IRanges_2.14.12 S4Vectors_0.18.3 BiocGenerics_0.26.0
loaded via a namespace (and not attached):
[1] drat_0.1.4 ggbeeswarm_0.6.0 colorspace_1.4-1 rprojroot_1.3-2 dynparam_1.0.0 htmlTable_1.13.1 XVector_0.20.0
[8] base64enc_0.1-3 rje_1.9 rstudioapi_0.9.0 farver_1.1.0 dynutils_1.0.2 bit64_0.9-7 ggrepel_0.8.0
[15] ranger_0.11.2 codetools_0.2-15 splines_3.5.1 knitr_1.22 Formula_1.2-3 jsonlite_1.6 cluster_2.0.7-1
[22] ggforce_0.1.3 shiny_1.2.0 readr_1.3.1 compiler_3.5.1 backports_1.1.3 assertthat_0.2.0 Matrix_1.2-14
[29] lazyeval_0.2.1 cli_1.1.0 later_0.7.5 tweenr_1.0.1 acepack_1.4.1 htmltools_0.3.6 tools_3.5.1
[36] igraph_1.2.4 gtable_0.2.0 glue_1.3.1 GenomeInfoDbData_1.1.0 reshape2_1.4.3 dplyr_0.8.0.1 Rcpp_1.0.1
[43] GA_3.2 iterators_1.0.10 DelayedMatrixStats_1.2.0 ggraph_1.0.2 xfun_0.5 stringr_1.4.0 ps_1.3.0
[50] testthat_2.0.1 mime_0.6 devtools_1.13.6 zlibbioc_1.26.0 MASS_7.3-50 scales_1.0.0 tidygraph_1.1.2
[57] babelwhale_0.0.0.9000 hms_0.4.2 promises_1.0.1 RColorBrewer_1.1-2 yaml_2.2.0 memoise_1.1.0 gridExtra_2.3
[64] dyndimred_1.0.0 rpart_4.1-13 latticeExtra_0.6-28 stringi_1.4.3 desc_1.2.0 foreach_1.4.4 checkmate_1.9.1
[71] shades_1.3.1 rlang_0.3.1 pkgconfig_2.0.2 bitops_1.0-6 lattice_0.20-35 purrr_0.3.2 patchwork_0.0.1
[78] htmlwidgets_1.3 bit_1.1-14 cowplot_0.9.4 pdist_1.2 processx_3.3.0 tidyselect_0.2.5 plyr_1.8.4
[85] R6_2.4.0 Hmisc_4.2-0 pillar_1.3.1 foreign_0.8-71 carrier_0.1.0 withr_2.1.2 units_0.6-2
[92] survival_2.42-6 RCurl_1.95-4.11 nnet_7.3-12 tibble_2.1.1 hdf5r_1.0.1 crayon_1.3.4 viridis_0.5.1
[99] grid_3.5.1 data.table_1.12.0 digest_0.6.18 xtable_1.8-3 tidyr_0.8.3 httpuv_1.4.5.1 munsell_0.5.0
[106] beeswarm_0.2.3 viridisLite_0.3.0 vipor_0.4.5
EDIT: I added the verbose=TRUE
result. Based on other issues I've seen it may have to do with the memory issues. Unfortunately, I cannot test it on the server due to docker restrictions, but I may try it out with a smaller data set in the future. If you have any other leads, fire away.
Hi @friedue This might be related to a memory issue. We do use sparse matrices internally now, but slingshot cannot handle these... Looking at the dimensions of your data, slingshot would use 6GB at max and run for 2 hours (as estimated by our guidelines app).
Could you perhaps first try to run this with a smaller dataset, so that we can be sure it is related to the dimensions of the dataset? Alternatively, you could also try to run it with paga
, which will use an estimated 2 GB of memory.
Thanks!
Hi,
I've tried my hands at slingshot via dynwrap and that didn't pan out.