Hello, I've read the help page and the vignette and I want to have the strand information of the probes used, but I can't see what the required format of the names of the vectors is. Can anyone provide a small example of how it's done ? It would be nice if there was a simple way to do this based on a data.frame or GRanges object of probe information, rather than having having to create lots of specially named vectors and mess around with environments. Also, the chr argument to segChrom has to be an integer vector, which means X and Y for human have to be artificially renamed as 23 and 24. The design could be more streamlined. - Dario. -------------------------------------- Dario Strbenac Research Assistant Cancer Epigenetics Garvan Institute of Medical Research Darlinghurst NSW 2010 Australia

Dear Dario please can you next time be clearer which package you are refering to. 'probeAnno' objects are defined in at least two packages: 'tilingArray' and 'Ringo'. (In R there is no assumption that the same class name can be used only by a single package in the world, this is why R has name spaces.) Assuming that you mean the tilingArray package, the 1st sentence of Section 2 in the vignette says: The function plotAlongChrom accepts an environment as its first argument, which is expected to contain objects of class segmentation with names given by paste(chr, c("+", "-"), sep="."), where chr is the chromosome identifier. So the strand is implied by the name of the segmentation object, e.g. "1.+" and "1.-" correspond to the Watson and Crick stands of chromosome 1 respectively. Regarding the design: the tilingArray package was written in 2005, 06. It predates GRanges, the oligo package, and almost anything else of that sort. The Ringo package was written later and corrected many of the less streamlined aspects of the initial attempt. The IRanges-infrastructure was developed more recently, and is by far much more elegant and performant. Personally, I do not see much value in refactoring the old tiling array code from the middle of last decade, when most people in the meanwhile have moved on to high-throughput sequencing, and the those who still use tiling arrays normally have lots of legacy code. However, you should feel free to do so, and you could contribute a tiling array package that you actually like :) Best wishes Wolfgang May/23/12 11:00 AM, Dario Strbenac scripsit:: > Hello, > > I've read the help page and the vignette and I want to have the > strand information of the probes used, but I can't see what the > required format of the names of the vectors is. Can anyone provide a > small example of how it's done ? It would be nice if there was a > simple way to do this based on a data.frame or GRanges object of > probe information, rather than having having to create lots of > specially named vectors and mess around with environments. Also, the > chr argument to segChrom has to be an integer vector, which means X > and Y for human have to be artificially renamed as 23 and 24. The > design could be more streamlined. > > - Dario. > > -------------------------------------- Dario Strbenac Research > Assistant Cancer Epigenetics Garvan Institute of Medical Research > Darlinghurst NSW 2010 Australia > -- Best wishes Wolfgang Wolfgang Huber EMBL http://www.embl.de/research/units/genome_biology/huber

> Assuming that you mean the tilingArray package, the 1st sentence of > Section 2 in the vignette says: > The function plotAlongChrom accepts an environment as its first > argument, which is expected to contain objects of class segmentation > with names given by paste(chr, c("+", "-"), sep="."), where chr is the > chromosome identifier. But there is only one probeAnno class, defined in Ringo, right ? Thanks for the reference to the vignette, that's what I needed to know. I assumed ?probeAnno was the best place to find out about creating an object. My understanding is that vignettes are to show how the package works on real data for an analysis, rather than defining how to use class constructors not documented in basic help files. Another aspect I found unclear was that the first two parameters to segChrom can be a matrix and a probeAnno object and how they were linked. Without reading the source code of segChrom, the user wouldn't know that the chrNumber.index items of a probeAnno object correspond to the row numbers/names of the intensity matrix. Which seems logical, but could be stated somewhere. I've made assumptions before about how public software works in the past, and it turned out it didn't work how I assumed, so I'm wary about ambiguities nowadays. > However, you should feel free to do so, and you could contribute a tiling > array package that you actually like :) Done it once before for my package - it was an adventure ! I think I might pass you up on your offer. - Dario.

Also worth describing in ?probeAnno > probes Error in validObject(object) : invalid class ?probeAnno? object: 1: Probe matches are not sorted (in increasing order) by their middle position on chromosome chr1.- and possibly others. invalid class ?probeAnno? object: 2: Probe matches are not sorted (in increasing order) by their middle position on chromosome chr1.+ and possibly others. invalid class ?probeAnno? object: 3: Probe matches are not sorted (in increasing order) by their middle position on chromosome chr10.- and possibly others. ... ... Also, I need to write a workaround for segChrom for any chromosomes that don't have probes on both strands (boutique design array). ... ... Running 'segment' on chromosome chr12.+ ... complete Running 'segment' on chromosome chr12.- ... complete Running 'segment' on chromosome chr13.+Error in probeAnno[w] : No mapping 'chr13.+.start' in this 'probeAnno' object. > probes A 'probeAnno' object holding the mapping between reporters and genomic positions. Chromosomes: chr1.- chr1.+ chr10.- chr10.+ chr11.- chr11.+ chr12.- chr12.+ chr14.- chr14.+ chr15.- chr15.+ chr16.- chr16.+ chr17.- chr17.+ chr18.- chr18.+ chr19.- chr19.+ chr2.- chr2.+ chr20.- chr20.+ chr21.- chr21.+ chr22.- chr22.+ chr3.- chr3.+ chr4.- chr4.+ chr5.- chr5.+ chr6.- chr6.+ chr7.- chr7.+ chr8.- chr8.+ chr9.- chr9.+ chrX.- chrX.+ chrY.- chrY.+ Microarray platform: Genome: Gives a valid object, so segChrom should work with it. - Dario

Sorry, bad example for the last part. Here's what I was meant to show. > segments <- segChrom(intensRBNorm, probes, chr = c("chr9", "chr10"), strands = c('+', '-'), nrBasesPerSegment = 500000, maxk = 500, step = 7) Running 'segment' on chromosome chr9.+ ... complete Running 'segment' on chromosome chr9.- ... complete Running 'segment' on chromosome chr10.+Error in segment(ychr, maxseg = nsegs, maxk = maxk) : maxseg must be an integer of length 1 between 1 and nrow(y)=0 > head(probes["chr10.+.start"]) integer(0) > head(probes["chr10.-.start"]) [1] 44872512 44872538 44872560 44872588 44872612 44872634 Only a few genes on chr10 - strand are tiled, none on + strand. I'll get around it by creating a probeAnno object for chromosomes that have both strands tiled, one for only +, and one for only -, and likewise for intensity matrices.