search

welch-lab / RcppPlanc

GNU General Public License v2.0
3 stars 1 forks source link

How to install RcppPlanc package on Ubuntu #3

Closed yinshanfang closed 7 months ago

yinshanfang commented 7 months ago

Hello, I wonder if anyone has the instruction of installation of RcppPlanc package on Ubuntu system. Thank you.

theAeon commented 7 months ago 
install.packages("RcppPlanc", repos = c(
  linux = 'https://welch-lab.r-universe.dev/bin/linux/jammy/4.3',
  sources = 'https://welch-lab.r-universe.dev',
  cran = 'https://cloud.r-project.org'))

should do it

mckaelaautumn commented 7 months ago

Hi, I'm still having an issue downloading RcppPlanc. I had no issue using rliger about a week ago, but have had issues today installing/loading the package in R on linux. This is my code df.liger <- OSTEO df.liger <- NormalizeData(df.liger, assay = "RNA", layer = "counts", save = "ligerNormData") df.liger <- FindVariableFeatures(df.liger) df.liger <- ScaleData(df.liger, split.by = "orig.ident", do.center = FALSE) df.liger <- scaleNotCenter(df.liger, assay = "RNA", save = "ligerScaleData", datasetVar = "orig.ident") df.liger <- runINMF(df.liger, k = 20, lamda = 5, datasetVar = "orig.ident", layer = "ligerScaleData", assay = "RNA", reduction = "inmf")

I get to here and get stuck here with error: "Package RcppPlanc is required for iNMF integration"

I have tried to above method and it seemed to work, but library(RcppPlanc) results in : Error: package or namespace load failed for ‘RcppPlanc’ in dyn.load(file, DLLpath = DLLpath, ...): unable to load shared object '/home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/RcppPlanc/libs/RcppPlanc.so': libhdf5_serial.so.103: cannot open shared object file: No such file or directory

I've also tried the recommended "devtools::install_github('welch-lab/RcppPlanc" and this is the error: Error: package or namespace load failed for ‘RcppPlanc’ in dyn.load(file, DLLpath = DLLpath, ...): unable to load shared object '/home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/00LOCK-RcppPlanc/00new/RcppPlanc/libs/RcppPlanc.so': /home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/00LOCK-RcppPlanc/00new/RcppPlanc/libs/RcppPlanc.so: undefined symbol: hwloc_topology_load Error: loading failed Execution halted ERROR: loading failed

Thank you!

theAeon commented 7 months ago

Seems like you don't have the dependencies installed for the r-universe binary. As you're using a conda environment, the r-universe binary won't work. I'm not sure that I've ever seen the compile succeed but loading fail b/c of hwloc symbols, b/c its supposed to compile it itself if it cant find the symbol.

Are you running this on an HPC cluster of some sort?

mckaelaautumn commented 7 months ago

It's weird because I installed rliger within my conda environment ~ a week ago and it worked great.

Yes, I'm on an HPC cluster.

theAeon commented 7 months ago

Well, we did just release a massive update. Can you run module spider hwloc? If theres a hwloc module, you may need to load it.

mckaelaautumn commented 7 months ago

My R version is 4.3.3 Seurat is 5.0.3 rliger is 2.0.0 and I installed blas, cmake, and hdf5 via conda.

Yeah, I'm having issues with the installation for modules based on permissions. Cannot sudo install. Thanks,

theAeon commented 7 months ago

I'm confused as to what you're trying to install re: modules. Does your HPC environment have a module command?

mckaelaautumn commented 7 months ago

running module spider hwloc gets: bash: module: command not found...

Tried to install, but getting permissions denied at make install step: install: cannot create directory ‘/usr/share/Modules’: Permission denied

I work within my labs server that's connected to the university HPC and my PI made it so we're supposed to create our own environments/load our own packages based on our needs. I am not the most tech savy and most of my experience is in R.

theAeon commented 7 months ago

I see-yeah, if you don't have a module command, definitely don't install it.

Can you see if there is a file containing "hwloc" in any one of /usr/lib, /usr/lib64, /lib, or /lib64?

mckaelaautumn commented 7 months ago

No I do not have hwloc in any of the lib directories...

theAeon commented 7 months ago

Huh. Can you get the output of devtools::install_github("welch-lab/RcppPlanc", force = TRUE, verbose = TRUE)?

if hwloc doesn't exist its...supposed to build it itself.

mckaelaautumn commented 7 months ago

** testing if installed package can be loaded from temporary location Error: package or namespace load failed for ‘RcppPlanc’ in dyn.load(file, DLLpath = DLLpath, ...): unable to load shared object '/home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/00LOCK-RcppPlanc/00new/RcppPlanc/libs/RcppPlanc.so': /home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/00LOCK-RcppPlanc/00new/RcppPlanc/libs/RcppPlanc.so: undefined symbol: hwloc_topology_load Error: loading failed Execution halted ERROR: loading failed

theAeon commented 7 months ago

the whole thing, please.

mckaelaautumn commented 7 months ago
Screenshot 2024-03-27 at 9 16 40 AM Screenshot 2024-03-27 at 9 18 31 AM Screenshot 2024-03-27 at 9 18 34 AM Screenshot 2024-03-27 at 9 18 38 AM
theAeon commented 7 months ago

fyi for the future-when you have a massive log like this you can upload it to https://gist.github.com/

makes it easier for everyone involved.

anyway, it looks like it is finding hwloc at /home/mckaelaautumn19/anaconda3/envs/seurat4/lib/libhwloc.so

can i get more info on your conda environment?

mckaelaautumn commented 7 months ago

For sure, sorry about that

Conda Package Installation conda update conda conda create -n seurat4 conda activate seurat4 conda config --remove channels conda-forge conda config --add channels conda-forge conda config --set channel_priority flexible conda install -c conda-forge r-xml conda install -c conda-forge r-devtools conda install -c conda-forge r-terra conda install -c conda-forge r-units conda install -c conda-forge r-sf conda install -c conda-forge r-spdep conda install -c conda-forge r-rgeos conda install -c conda-forge screen conda install -c conda-forge r-nloptr conda install -c conda-forge r-Cairo conda install -c conda-forge r-lme4 conda install -c conda-forge r-ggrastr conda install -c conda-forge r-BiocManager conda install -c conda-forge r-seurat conda install -c conda-forge r-SeuratObject conda install -c conda-forge r-ggplot2 conda install -c conda-forge r-plotly conda install -c conda-forge r-dplyr conda install -c conda-forge r-patchwork conda install -c conda-forge r-harmony conda install bioconda::bioconductor-glmgampoi conda install -c conda-forge r-R.utils conda install -c conda-forge r-SingleCellExperiment conda install -c conda-forge r-monocle3 conda install -c conda-forge r-SeuratWrappers conda install bioconda::r-liger conda install cmake conda install blas

theAeon commented 7 months ago

okay-I'm testing this on my end, bear with.

there are two possibilities I'm seeing here. One of them is that I messed up the bit of the cmakelists file that links an existing hwloc with RcppPlanc and the other is that conda's build is weird.

I'm leaning towards the former.

mckaelaautumn commented 7 months ago

Thank you so much! I only just recently was pointed towards using rliger and I was getting excellent results from applying it to my data so very excited to continue using it!

KaiWaldrant commented 7 months ago

From what I can tell I also seem to have an issue with hwloc when trying to install the package:

https://gist.github.com/KaiWaldrant/439ac19c88264e968a469a7e26d5d500

sessioninfo:

> sessionInfo()
R version 4.3.3 (2024-02-29)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 22.04.1 LTS

Matrix products: default
BLAS:   /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.10.0 
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0

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

time zone: UTC
tzcode source: system (glibc)

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

loaded via a namespace (and not attached):
 [1] compiler_4.3.3    assertthat_0.2.1  R6_2.5.1          Matrix_1.6-5     
 [5] cli_3.6.2         Rcpp_1.0.12       reticulate_1.35.0 bspm_0.5.5.1     
 [9] grid_4.3.3        anndata_0.7.5.6   jsonlite_1.8.8    rlang_1.1.3      
[13] png_0.1-8         lattice_0.22-5

edit: Not using conda

theAeon commented 7 months ago

From what I can tell I also seem to have an issue with hwloc when trying to install the package:

Stupid mistake on my part w/ the CMakeLists file. Should be fixed.

theAeon commented 7 months ago

Thank you so much! I only just recently was pointed towards using rliger and I was getting excellent results from applying it to my data so very excited to continue using it!

Alright-I definitely found the issue, but I'm not entirely sure if I fixed it. Try installing it again and if it fails get me the lines following Linking CXX Shared library RcppPlanc.so

KaiWaldrant commented 7 months ago

It seems to have been fixed.

mckaelaautumn commented 7 months ago

Hi, so installing RcppPlanc is working, but library(RcppPlanc) gets error: Error: package or namespace load failed for ‘RcppPlanc’ in get(Info[i, 1], envir = env): lazy-load database '/home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/RcppPlanc/R/RcppPlanc.rdb' is corrupt In addition: Warning message: In get(Info[i, 1], envir = env) : internal error -3 in R_decompress1

theAeon commented 7 months ago

Hi, so installing RcppPlanc is working, but library(RcppPlanc) gets error: Error: package or namespace load failed for ‘RcppPlanc’ in get(Info[i, 1], envir = env): lazy-load database '/home/mckaelaautumn19/anaconda3/envs/seurat4/lib/R/library/RcppPlanc/R/RcppPlanc.rdb' is corrupt In addition: Warning message: In get(Info[i, 1], envir = env) : internal error -3 in R_decompress1

Quick google says that restarting your R session should take care of that one.

mckaelaautumn commented 7 months ago

Great! It's working, Thank you so much! I did have a momentary warning OpenBLAS Warning : Detect OpenMP Loop and this application may hang. Please rebuild the library with USE_OPENMP=1 option.

but resolved with conda install conda-forge::openmp Thanks again!

theAeon commented 7 months ago

Good instinct. openblas can have issues when built with pthreads while openmp is enabled. I try to mitigate the damage of that by forcing openblas down to 1 thread when this situation is detected, so it wouldnt have been a huge issue, but replacing it with conda-forge's openmp build is definitely better for performance!