[BioC] Wow, what have I done

[Kasper Daniel Hansen]

On Dec 10, 2007, at 6:26 PM, Loren Engrav wrote:

> Thank you
>
> Am using G5
> And I have Leopard but have not installed as "too little hard drive  
> space"
> And have 32bit R version 2.6.1 (2007-11-26) which I run with R.app  
> and is
> fine but fails to ReadAffy() 81 .cel files (cannot make vector...)
>
> So am now trying for G5 ppc 64bit R to read 83 .cel files
>
> I went to page that says...
>
> 64-bit R for Mac OS X
> All binaries on this page are strictly experimental. For licenses see
> individual projects (basically covered by GPL). All downloads are  
> subject to
> acceptance of the disclaimer below. Feedback is welcome, but please  
> don't
> use R-bugs as this is not an official release!
> R for PowerPC 64-bit (G5) on Mac OS X 10.4 (Tiger)
>
> And obtained the
>
> Monster package of 64-bit R
> The following tar-ball includes R 2.3.1 + X11 client + binaries of  
> all CRAN
> packages that build cleanly + binaries of default Bioconducductor  
> packages
> that build cleanly. Packages that require external libraries that  
> are not
> part of Tiger are not included.
> Download R-2.3.1-ppc64-monster.tar.gz (ca. 530MB!!)
>
> And installed same and it ReadAffy() 81 cel files very nicely but
> BiocLite fails for some packages and affyQCReport and simpleaffy  
> are not
> included

Yes, all of this was reasonably clear from your earlier description.

> So
> If this monster package of 2.3.1 is bad out of date, is there a  
> monster
> package of 2.6.1 64bit for G5 ppc?

No. And this monster package you are using is more than 1.5 years out  
of date.

The only way to get something current which everyone will recommend  
(essentially if you have any problems using such an old version of R  
people will tell you to upgrade before they help), is to do what I said:
1) Install Leopard
2) Install R-2.6.1 from the package on this page
http://r.research.att.com/
You probably want 2.6.1 (instead of the development version 2.7.0)  
and you most certainly want one which under "status" says "ppc64". As  
you see on the package, Leopard is required.
3) You then need to be able to install packages "from source". This  
includes getting Xcode-3, and tools from
http://r.research.att.com/tools/
and possibly other stuff.
4) Then set
option(pkgtype = "source")
and do the usual
biocLite()

Kasper





>> From: Kasper Daniel Hansen <khansen at stat.Berkeley.EDU>
>> Date: Mon, 10 Dec 2007 17:57:55 -0800
>> To: Loren Engrav <engrav at u.washington.edu>
>> Cc: "bioconductor at stat.math.ethz.ch" <bioconductor at stat.math.ethz.ch>
>> Subject: Re: [BioC] Wow, what have I done
>>
>> On Dec 10, 2007, at 5:29 PM, Loren Engrav wrote:
>>
>>> So I did and this is half Mac list half Bio list
>>>
>>> curl -O http://r.research.att.com/exp/R-2.3.1-ppc64.tar.gz
>>> sudo tar fvxz R-2.3.1-ppc64.tar.gz -C /
>>> sudo ln -sf /usr/local/lib64/R/bin/R /usr/local/bin/R64
>>>
>>> As per <http://cran.fhcrc.org/>
>>>
>>> And cool, I have 64 bit R which launches
>>
>> But your R version is 2.3.1 - seriously outdated.
>>
>> You should be using the build of R-2.6.1 that says "leopard only,
>> pp64" on
>> http://r.research.att.com/
>> You need to be using leopard
>>
>> (I hope you are using a G5 cpu, otherwise it will not work, I can see
>> from earlier posts that you are using a PowerPC cpu).
>>
>> Then you need to do
>>   options(pkgType = "source")
>> and set up your mac to install bioconductor from source (otherwise
>> you will not get 64bit versions of the packages, which is what you
>> want). This may be painful if you have not done so before. Them you
>> can do
>>   biocLite()
>>
>> It is possible as far as I know to get 64bit R to run under Tiger,
>> but it is supposedly very painful. The stuff above requires you to
>> upgrade to Leopard however.
>>
>> Kasper
>>
>>
>>> But the
>>>
>>> source("http://bioconductor.org/biocLite.R")
>>> biocLite()
>>>
>>>  method of installing packages failed with complicated error  
>>> messages
>>>
>>> So I obtained the monster 64bit package from <http:// 
>>> cran.fhcrc.org/>
>>>  and dragged it all into the various /usr folders
>>>
>>> And super,
>>> ReadAffy now reads 81 .cel files no problem with 2.5 gig
>>> So the 64bit is far more efficient than the 32bit
>>>
>>> But now 3 new questions
>>> 1) does the biocLite method of packages work with R64bit
>>> 2) affyQCReport is not in the monster, how do I get affyQCReport
>>> into 64 bit
>>> 3) when quitting R in the Terminal (Mac) I save the image but do
>>> not know
>>> where it went nor what it is named, where is it, what is it called
>>>
>>> Thank you
>>>
>>> -- 
>>> Loren Engrav
>>> Univ Washington
>>> Seattle
>>>
>>>> From: "Marcus G. Daniels" <mdaniels at lanl.gov>
>>>> Date: Mon, 10 Dec 2007 12:15:53 -0700
>>>> To: "bioconductor at stat.math.ethz.ch"  
>>>> <bioconductor at stat.math.ethz.ch>
>>>> Subject: Re: [BioC] Wow, what have I done
>>>>
>>>> Loren Engrav wrote:
>>>>> 3) try 64bit R
>>>>>
>>>> Also if you are a Linux user, you can try VMware Fusion together  
>>>> with
>>>> some x86_64 distribution of Linux (e.g. Fedora 8 x86_64 has a 64
>>>> build
>>>> of R). This will get you up past the 4GB software limit (up to 8GB
>>>> virtual memory).
>>>>
>>>> _______________________________________________
>>>> Bioconductor mailing list
>>>> Bioconductor at stat.math.ethz.ch
>>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>>> Search the archives:
>>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>>
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>> Search the archives: http://news.gmane.org/
>>> gmane.science.biology.informatics.conductor
>>
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/ 
> gmane.science.biology.informatics.conductor