Bioconductor not working through Unix command line

Question

[k.e.pendleton]

I'm working through the Bioconductor book and am currently int he RNA-Seq statistics: R and Bioconductor section. I've actually used both R and Bioconductor before, but in Windows, not Linux.

I have R installed. If I try to run conda install r again, everything freezes up and I have to open up another window and sudo kill the process.

I did the following:

$ conda install -y bioconductor-deseq bioconductor-deseq2 bioconductor-edger r-gplots
Fetching package metadata .................
Solving package specifications: .

# All requested packages already installed.
# packages in environment at /home/moltres/miniconda3/envs/bioinfo:
#
bioconductor-deseq        1.24.0                 r3.3.2_0    bioconda
bioconductor-deseq2       1.16.1                 r3.3.2_0    bioconda
bioconductor-edger        3.16.5                 r3.3.2_0    bioconda
r-gplots                  2.17.0                 r3.3.2_0    bioconda
(bioinfo) 
$ curl -O http://data.biostarhandbook.com/rnaseq/code/deseq1.r
$ curl -O http://data.biostarhandbook.com/rnaseq/code/counts.txt
$ cat counts.txt | Rscript deseq1.r 3x3 > results_deseq1.txt
Error in library(DESeq) : there is no package called 'DESeq'
Execution halted

So even though I just downloaded DESeq, it says I don't have the DESeq package. I did check my R version, and it's 3.4.3. I suspect that could be a problem, but I don't know how to change it in Linux. When I try doing everything in RStudio, I get a bunch of errors in dependencies and configuration upon using biocLite("DESeq").

Any ideas what's going wrong? I would appreciate any help.

[Philipp Bayer]

What is the output of which Rscript? I have a hunch that the Rscript you're using there is not the Rscript from anaconda

[k.e.pendleton]

$ which Rscript
/home/moltres/miniconda3/envs/bioinfo/bin/Rscript

Okay. How do I know if it's from anaconda? Is it in the wrong directory? Thanks for the response. Still trying to figure this out.

Also, if it counts for anything I tried to uninstall everything. I did...

$ sudo apt-get --purge remove r-base
$ sudo apt-get --purge remove r-base-dev
$ sudo apt-get --purge remove r-base-core

I also went into /home/moltres/miniconda3/envs/bioinfo/bin/Rscript and deleted the R and RScript files. I tried to re-download R using conda install r.

It's still getting stuck and not downloading. I'm not really sure what else to do at this point...

[Philipp Bayer]

Hm that's the correct Rscript -
By 'it's still getting stuck and not downloading', what do you mean? Anaconda is not doing anything? The R installation is big and may take a long while to resolve dependencies in ananonda

Edit: An idea: Have you tried installing DESeq and DESeq the non-Anaconda way?

Start R using 'R', then copy paste:

source("https://bioconductor.org/biocLite.R")
biocLite('DESeq')
biocLite("DESeq2")
biocLite("edgeR")

That will install DESeq, DESeq2 into your anaconda-installed R, maybe that'll do it?

I tried to reproduce your error but I'm not even getting that far :D I'm getting heaps of segmentation faults haha

[k.e.pendleton]

So, it turns out that I'm just impatient, and R did download. I have good and bad news.

When I try to run the Rscript in command line, it still fails.

    $ cat counts.txt | Rscript deseq1.r 3x3 > results_deseq1.txt
bash: /usr/bin/Rscript: No such file or directory
(bioinfo) 

$ Rscript
bash: /usr/bin/Rscript: No such file or directory
(bioinfo) 

$ which Rscript
/home/moltres/miniconda3/envs/bioinfo/bin/Rscript
(bioinfo) 

$ which R
/home/moltres/miniconda3/envs/bioinfo/bin/R
(bioinfo)

However, when I open RStudio, I can now install DESeq, and do library(DESeq) without getting error telling me that the directory is wrong. However, when I try to open the deseq1.r script, it tells me "No such file or directory." However, I checked the working directory, and the file is in the same folder as my working directory. I tried to look up this error, and it told me to save it with encoding UTF-8 encoding and re-open it. Tried it. Didn't work. I can't even open a new script window--it won't let me type anything. Do you know of any way to troubleshoot this?

[k.e.pendleton]

Just for future reference if anyone ever has this issue, I've mostly fixed it now. For some reason, bash thinks R is in one place, even though it also knows it's in another. I don't really get it.

ANYWAY.

If I run the command where it specifies Rscript in the correct place:

$ cat counts.txt | /home/moltres/miniconda3/envs/bioinfo/bin/Rscript deseq1.r 3x3 > results_deseq1.txt

Then this works perfectly. I get the output file.

I've also uninstalled Rstudio, and then reinstalled it using

$ conda install rstudio

The same way I did when I re-downloaded R.

If I want to run rstudio, just typing rstudio in command line doesn't work. I had to type in the full location again.

HOWEVER...if I closed my terminal, then opened a new terminal and ran

$ source activate bioinfo

again, it started working if I just typed rstudio.

So now R seems to be cooperating with everyone else, and they're all a big happy family trying to teach me bioinformatics. And also patience. Unfortunately, I'm not great at either.

Thanks for the help, everyone!

[Philipp Bayer]

For some reason, bash thinks R is in one place, even though it also knows it's in another. I don't really get it.

bash looks at the environment variable PATH, whichever comes first in PATH is used first, so if you have

PATH=/usr/bin/:/home/kependleton/anaconda/bin

then bash will use the system R. You can check this using echo $PATH.

A workaround is to flip the order, or to put the anaconda one first:

Either:

export PATH=/home/kependleton/anaconda/bin/:/usr/bin/

or:

export PATH=/home/kependleton/anaconda/bin/:$PATH

[k.e.pendleton]

Ah. Thanks! I'll try to keep this in mind. I struggle looking for solutions to some easy issues since I don't have the right language yet to properly say what's going wrong.

Answer 1

[Arindam Ghosh]

I am not sure about the way you are using it. But I use the tools within R. The packages were installed within R as mentioned in the Bioconductor support page here. For doing any statistical analysis I write an Rscript and execute in terminal or do it one by one within R terminal. And I thing to remember I need to load each libraries using library() function.

[WouterDeCoster]

It should work, if as Philipp Bayer mentioned the Rscript is the one in anaconda/bin, which it probably isn't. Changing the order of $PATH may help. As you can see from the error message the Rscript uses library(DESeq), so that part is alright.