James W. MacDonald wrote: > Gregor GORJANC wrote: > >> Hello! >> >> Can someone point me to explanation/documentation why Bioconductor >> uses also some other "package installation" design in contrast to R. I >> appologize for ignorance, but I am just starting with Bioconductor and >> was surprised on this difference to pure R. >> > I think there are at least two reasons. First, the reposTools/getBioC > paradigm allows one to specify a 'theme' of packages to download, based > on the sort of analyses one might be interested in performing. For > instance, getBioC("affy") will get all the packages you will likely need > to use for Affymetrix chip analyses. AFAIK, there is no such > functionality in the standard install.packages(). Instead, you have to > seek out the packages you want to install from CRAN. > > Second, rather than having all the packages in one big repository (a la > CRAN), the BioC repositories are separated out by function or type, so > installing a set of packages may require iteration through 11 different > repositories in order to get all the requested packages. I don't believe > that this was easy to do with the base installation functionality when > Bioconductor was first started. > Jim, > > Jim, thanks for the response. But that would also be usable for "pure R". Why wasn't this just merged together. -- Lep pozdrav / With regards, Gregor Gorjanc ---------------------------------------------------------------------- University of Ljubljana Biotechnical Faculty URI: http://www.bfro.uni-lj.si/MR/ggorjan Zootechnical Department mail: gregor.gorjanc <at> bfro.uni-lj.si Groblje 3 tel: +386 (0)1 72 17 861 SI-1230 Domzale fax: +386 (0)1 72 17 888 Slovenia, Europe ---------------------------------------------------------------------- "One must learn by doing the thing; for though you think you know it, you have no certainty until you try." Sophocles ~ 450 B.C.

> package managing system, due to dependency issues. I understand that > dependencies might be a bit more complicated than in R, but I'am sure that > such improvements would be of much more benefit if they would also be used > by "pure R". Therefore, improvements of R's related projects as > Bioconductor and others should also find their way "back" into R. Finally, > I realize that usually their is lack of people to do all this. In a sense, they have. People such as Fritz Leisch and Kurt Hornik (who work hard on CRAN and (as I understand it) develop a lot of the R based package management software) were involved in the design process of reposTools (the Bioconductor package management package). Also, while it hasn't been a direct transfer of code into R, many of the same/similar ideas are what found there way over into R's code base with the benefit of having a lot of concepts tested to see what works, what doesn't work, etc. As an aside, one issue preventing any sort of direct code transfer is that reposTools relied on S4 classes - which at least at the time (and I assume is still the case) is not what is being used in package management code for R.

Of the 16 members of R core, 5 are members of Bioconductor core. I do not think that you need to worry that there is no communication. Many changes in R in the last years have been initiated because of Bioonductor. And according to Seth (I have not yet looked at it myself), R's installation routines have been updated quite a bit in the new release. Probably at least partly inspired by Bioconductor. Kasper On Fri, Apr 29, 2005 at 01:04:34PM +0200, Gregor GORJANC wrote: > Seth thanks for clarification. Now I understand the "whole" thing. However, > I would like to add a comment. Don't take me wrong. A appreciate very much > work done by Bioconductor, R team and other volunteers. All of them are > just doing marvelous job. > > As I understand Seth's comments, Bioconductor developed new/improved > package managing system, due to dependency issues. I understand that > dependencies might be a bit more complicated than in R, but I'am sure that > such improvements would be of much more benefit if they would also be used > by "pure R". Therefore, improvements of R's related projects as > Bioconductor and others should also find their way "back" into R. Finally, > I realize that usually their is lack of people to do all this. > > Thanks again and good luck! > > >To follow up on James' comments, when Bioconductor started, R's > >install.packages and friends did not have a way of handling > >dependencies between packages. > > > >Bioconductor was one of the first collections of packages that had a > >complicated/deep dependency structure. In many ways this is a good > >thing as it demonstrates code reuse and collaboration. The project > >had to "roll its own" package installation code in order to deal with > >the dependencies. > > > >In the latest release of R, the package installation tools have been > >vastly improved. They are now capable of resolving dependencies > >between a collection of package repositories. > > > >The Bioconductor team is evaluating to what extent we can utilize the > >new capabilities built into R and move away or adapt the reposTools > >code. > > > >I think we will continue to have an install script ala getBioC in > >order to simplify the process and provide some themed installs, as > >James mentioned. > > > >Hope that helps a bit for some background. > > -- > Lep pozdrav / With regards, > Gregor Gorjanc > > ---------------------------------------------------------------------- > University of Ljubljana > Biotechnical Faculty URI: http://www.bfro.uni- lj.si/MR/ggorjan > Zootechnical Department mail: gregor.gorjanc <at> bfro.uni-lj.si > Groblje 3 tel: +386 (0)1 72 17 861 > SI-1230 Domzale fax: +386 (0)1 72 17 888 > Slovenia, Europe > ---------------------------------------------------------------------- > "One must learn by doing the thing; for though you think you know it, > you have no certainty until you try." Sophocles ~ 450 B.C. > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor -- Kasper Daniel Hansen, Research Assistant Department of Biostatistics, University of Copenhagen