The best way to do this is without a doubt to use R-3.3.0 beta and use biocLite, as Martin describes here. This way you get the latest version of minfi which also contains some bug fixes related to EPIC arrays, which you don't get it you install the packages from Bitbucket in (say) R.3.2.x.
Have another question and I have to apologise if this sounds like a silly question to ask (new to all of this)
So I've installed the two packages so does that mean when I run minfi it will automatically use the EPIC manifest and anno or will it use the 450k ones? Is there a command I need to use to run the EPIC files in R?
Also I'm going through the minfi tutorial using the minfiData package and I guess its written up for the 450K array. So just wanted to know again when I run minfi using my data files will it use the EPIC manifest etc. I am a bit confused because in the tutorial the RGchannel set is set for the 450K.
What version of R are you using? If you are using R-3.2.z, for some value of z? Then you need to use R-3.3.0. Start R-3.3.0, make sure that BiocInstaller is not installed (say remove.packages("BiocInstaller") until an error indicating that BiocInstaller was not found), then source("https://bioconductor.org/biocLite.R").
Normally you would install it with biocLite("IlluminaHumanMethylationEPICmanifest").
On Windows, in devel, and at this precise moment in time, there are important CRAN and Bioconductor packages that are not available in 'binary' format, and require that you have the Rtools compiler available to install these from source (and even then, because they are not available as binaries suggests that there are special challenges to compiling them).
Maybe you have access to a linux or mac, where it will be more straight-forward? (again, this is only relevant to using Bioc-devel, on Windows, at this particular moment in time; likely a week from now biocLite() will just work for this package on all platforms).
Although they primarily contain annotation rather than code, these can be treated like R software packages. Hopefully at some point in the future they will be included in Bioconductor and you will be able to use biocLite() to install them.
However, for now you can install straight from the downloaded tar.gz file. If you use RStudio then you can do:
Tools ->
Install Packages ->
Install From "Package Archive File (tar.gz)" ->
Select the tar.gz file
If you are working at the command line then you can run something like the following in a terminal:
R CMD INSTALL /path/to/IlluminaHumanMethylationEPICmanifest_0.3.0.tar.gz
Hi Mike, I've followed the steps to pointed above after downloading IlluminaHumanMethylationEPICanno.ilm10b2.hg19_0.6.0.tar.gz. Unfortunately, the installation process throws these errors :
Have another question and I have to apologise if this sounds like a silly question to ask (new to all of this)
So I've installed the two packages so does that mean when I run minfi it will automatically use the EPIC manifest and anno or will it use the 450k ones? Is there a command I need to use to run the EPIC files in R?
Also I'm going through the minfi tutorial using the minfiData package and I guess its written up for the 450K array. So just wanted to know again when I run minfi using my data files will it use the EPIC manifest etc. I am a bit confused because in the tutorial the RGchannel set is set for the 450K.
If you are using minfi version 1.17.10 or greater it should all just work. But it seems it is not available through Bioc for Windows right this moment.
The best way to do this is without a doubt to use R-3.3.0 beta and use biocLite, as Martin describes here. This way you get the latest version of minfi which also contains some bug fixes related to EPIC arrays, which you don't get it you install the packages from Bitbucket in (say) R.3.2.x.
Hi guys,
Have another question and I have to apologise if this sounds like a silly question to ask (new to all of this)
So I've installed the two packages so does that mean when I run minfi it will automatically use the EPIC manifest and anno or will it use the 450k ones? Is there a command I need to use to run the EPIC files in R?
Also I'm going through the minfi tutorial using the minfiData package and I guess its written up for the 450K array. So just wanted to know again when I run minfi using my data files will it use the EPIC manifest etc. I am a bit confused because in the tutorial the RGchannel set is set for the 450K.
Thanks again
Hi Martin,
That for your suggestion. I installed bioc but it seems to be v3.2. Do you know where i can find bioc 3.3?
Rasi
What version of R are you using? If you are using R-3.2.z, for some value of z? Then you need to use R-3.3.0. Start R-3.3.0, make sure that BiocInstaller is not installed (say
remove.packages("BiocInstaller")
until an error indicating that BiocInstaller was not found), thensource("https://bioconductor.org/biocLite.R")
.I managed to install Bioc 3.3 but now when I try to install IlluminaHumanMethylationEPICmanifest i get the following error:
Warning messages:
1: running command '"C:/R-33~1.0BE/bin/x64/R" CMD INSTALL -l "C:\R-3.3.0beta\library" C:\Users\kanis\AppData\Local\Temp\Rtmpa0qDCj/downloaded_packages/IlluminaHumanMethylationEPICmanifest_0.3.0.tar.gz' had status 1
2: In install.packages(pkgs = doing, lib = lib, ...) :
installation of package ‘IlluminaHumanMethylationEPICmanifest’ had non-zero exit status.
so I dont think it has installed properly. What do you think is going wrong?
Rasi
Normally you would install it with biocLite("IlluminaHumanMethylationEPICmanifest").
On Windows, in devel, and at this precise moment in time, there are important CRAN and Bioconductor packages that are not available in 'binary' format, and require that you have the Rtools compiler available to install these from source (and even then, because they are not available as binaries suggests that there are special challenges to compiling them).
Maybe you have access to a linux or mac, where it will be more straight-forward? (again, this is only relevant to using Bioc-devel, on Windows, at this particular moment in time; likely a week from now biocLite() will just work for this package on all platforms).