[BioC] update annotations without updating bioconductor

[Marc Carlson]

Hi Robert,

There is a good chance that you will have to (at a minimum) update
AnnotationDbi just to pull this off.  That said, I still think it's
probably a bad idea.  Each release, the group of annotations we put out
is designed as a "set", so if you upgrade just one annotation package,
you really HAVE to upgrade all of them.  So for example, if you were
using hgu95av2.db, and you upgraded that, you would also need to upgrade
GO.db, KEGG.db, and org.Hs.eg.db just to be able to use the hg95av2.db
package.  And if you then went to use a chip package from mouse, (say
moe430a.db), then you would also have to update that (because earlier
you had updated GO.db and KEGG.db), along with the org.Mm.eg.db
package...  And so on.  There could also be problems with packages like
GOstats or any other packages that make heavy use of annotations (and
there are MANY of these), that could reasonably be expecting things to
work in a certain way within a certain release.

Hopefully from these descriptions you can see why mixing and matching is
probably a bad idea?  You have to really know a lot about what is
happening behind the scenes in order to pull it off, and even then, it's
pretty dicey.  I have a very good handle on how the annotations have
changed over the years, and I have tried very hard to minimize backwards
compatible impacts on how they have changed, and yet, I would never
intentionally do this.  So I can't in good conscience recommend that you
do it either. 

There are very good reasons why we recommend for people to use
biocLite() and update.packages().  It is the only way we can assure you
that we have tested that these things at least should work correctly
together for you.  When you mix packages from different releases, you
are basically running code in a way that has not been tested. 


  Marc



On 11/05/2010 06:34 AM, James W. MacDonald wrote:
> Hi Robert,
>
> It's always possible to mix and match package versions - a significant
> portion of the noise on this list consists of people admonishing
> posters to use biocLite() rather than downloading and installing things.
>
> So you can always download and install the updated annotation
> packages, and they may well work OK (I don't think Marc made any
> changes that are not backward-compatible). However, you are on your
> own if you have any odd results. Caveat emptor.
>
> Best,
>
> Jim
>
>
>
> On 11/5/2010 9:18 AM, Robert M. Flight wrote:
>> Hi All,
>>
>> I'm wondering if it is possible to update annotation packages without
>> updating the core bioconductor itself. For example, if I wanted to
>> update various "org.*.db" annotation packages, GO.db, and KEGG.db
>> without updating R and Bioconductor, how would I do that? I do plan to
>> update R and Bioconductor, but the system I am using and some of the
>> packages I am using took a lot to get working correctly, and I would
>> like to enable updated annotations while taking my time to upgrade R
>> and Bioconductor, without worrying too much about breaking something.
>>
>> Currently using R 2.11 and Bioconductor 2.6.
>>
>> Thanks in advance,
>>
>> Robert
>>
>> Robert M. Flight, Ph.D.
>> University of Louisville Bioinformatics Laboratory
>> University of Louisville
>> Louisville, KY
>>
>> PH 502-852-0467
>> EM robert.flight at louisville.edu
>> EM rflight79 at gmail.com
>>
>> Williams and Holland's Law:
>>         If enough data is collected, anything may be proven by
>> statistical methods.
>>
>> _______________________________________________
>> 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
>