We come out with new idea of normalization and tested on Affy U95 serial dilution data set. As a result of the normalization, we got a matrix: row as probe (PM intensity) (the probe of the same probe set are adjacent and if there are more than 16 probe sets, we only keep the first 16); column as different chips. Now we want to use BioC to do the further analysis in order to test the specificity and sensitivity of our methods. However, we cannot figure out how to construct the AffyBatch object from our matrix. Please help us to solve the problem. Thanks. L

Well first a note: an AffyBatch object contains one row per probe on the chip (and I mean _all_ probes: ncol * nrow). It uses the row number as a link to (x,y) position. It is therefore quite important that the object has the correct number of rows. The link goes a bit like this gene -> probeSet -> set of (x,y) coordinates -> row numbers If you delete eg. all control probes from the AffyBatch object the last step will be messed up. Now, I am not sure whether you have done summarization. If that is the case you should have a matrix with one row per gene, with the rownames equal to the affy probeset identifiers. If you are not at this step, you probably want to plugin your method into the affy package set of routines. This is quite doable, but you need to spend quite some time on the documentation. I think the vignette explains it (ok, perhaps not a complete walkthorugh, but at least points you in the right direction). If you look at the help files for "exprsSet", "phenoData" and "AffyBatch" object (and you may need to access them using class?AffyBatch for instance) you should find examples of creating AffyBatch objects. The key (or most difficult part) is actually creating the phenoData slot. You can also find a sample way of generating an AffyBatch object by looking at the last lines of the code for the read.affybatch function. I suggest you spend some time reading the help pages referred to above. And then return if you have more questions. Kasper On Aug 26, 2005, at 7:55 AM, Lizhe Xu wrote: > We come out with new idea of normalization and tested on Affy U95 > serial dilution data set. As a result of the normalization, we got a > matrix: row as probe (PM intensity) (the probe of the same probe > set are adjacent > and if there are more than 16 probe sets, we only keep the first > 16); column > as different chips. Now we want to use BioC to do the further > analysis in > order to test the specificity and sensitivity of our methods. > However, we > cannot figure out how to construct the AffyBatch object from our > matrix. > Please help us to solve the problem. Thanks. > > > L > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor >