Thank you very much Sean, I have been working with function aggregate and it is exactly what I need. However, there is still a painful detail that I cannot get rid of. I hope you can help me too with this. I have this text file: A B C D d1 2 23 2 d1 4 22 2 d1 5 24 2 d2 10 7 2 d2 20 8 3 d1 7 23 2 d3 2 14 30 d3 4 14 50 d2 30 8 4 d4 12 13 15 d5 1 5 90 d2 40 7 3 d6 34 2 5 (I use it as a test) If I type: > data<-read.table("test.txt",sep="\t") > agr<-aggregate(data[2:4], by=list(data$V1), FUN=mean) I get 21 warning messages and all the values are "NA", including header B, C, and D. However, if I remove A,B,C,D from the previous file, and type the same commands, it works perfectly fine, getting what I wanted. The problem is that the real datasets I need to work with are really large and it is difficult to remove and add the headers without danger of doing something wrong. Is there any command or parameter that I should introduce to the function in order to solve this issue? Thank you so much, Hernando 2010/6/11 Sean Davis <sdavis2@mail.nih.gov> > > > On Fri, Jun 11, 2010 at 8:12 AM, Hernando Martínez <hernybiotec@gmail.com>wrote: > >> Hello everyone, my name is Hernando, and I am new to R. I have a little >> problem that maybe you can help me with, as I have been looking through >> the >> packages with no success, and it shouldn't be very difficult to solve. >> I have a text file containing a list of genes, with expression values for >> each along a set of microarray experiments. Ex: >> >> geneID sample1 sample 2 .... >> >> gene1 45 58 .... >> >> gene1 43 63 ..... >> >> gene2 32 21 .... >> >> ...... ..... ...... ..... >> >> In this list, there are some genes repeated, but with different values >> (like >> in the example). This repetitions come from different probes targeting the >> same gene. >> What I want is a new text file, but with each gene appearing only once, >> and >> with three possibilities for the expression values of repeated genes: >> >> - Each value (for each column (sample)) is the average of the previous >> values (in the example, sample 1 for gene1 should be 44, and 60,5 in >> sample >> 2) >> - Instead of the average, the median. >> - The highest values. >> >> I would prefer the median or the average, but I don't know if getting the >> highest values is easier. >> >> I have seen this function: "findLargest" of "genefilter" package, but it >> works with probes and I have already converted files (to geneIDs). >> >> > Hi, Hernando. Have a look at the aggregate() function. > > Sean > > -- Hernando Martínez Vergara [[alternative HTML version deleted]]