Hi all, I'm using a nylon technology, with a radioactive labelling of the sample, and with cDNA probes spotted on the membrane. We do first an hybridisation with a vector oligonucleotide, that match all spot on the chip. This is to adjust for the different quantities spotted on the chip. Then we hybridize with the complex set issued by RTPCR from the sample. I can't figure out really what it implies when i use the different methods in Bioc. As it is a single channel procedure i'd think it's closer to affy chips for example. But in the other hand, as we do the first hybridization, and as it is a cDNA technology, i wrote a read function to load my data in a RGlist object, and then use it in limma. In this function, R channel correspond to the complex set, and G channel to the vector set. (two successive hybridisations on the same membrane). But in fact my ratio "complex / vector" has not really the same meaning as "sample labelled with Cy5 / reference sample labelled with Cy3". These questions raised as i was reading the monograph on bioconductor. The methods used to identify differentially expressed genes are not the same for an affybatch, and for a batch using limma for example. As i'm a wet-lab biologist, and as my skills in statistics are low, i don't really understand if it's the same or not, but i would say it's not ? So did you use different approaches because in one case your are using a signal from a single channel, oligo chip, and on the other one, a dual channel with a ratio of expression of a cDNA chip ? If it is true, the fact that my ratio "doesn't really represent the same thing as a dual channel cDNA chip" would mean that i have to use a specific method, and then which is the best. I don't know if what i say is clear, but i would appreciate your comments on this. Do you think in this case i should consider my ratio as a normalised (a first step of) value, and then think as a single channel ? Thanks for any help, Julien

Hi all, I'm using a nylon technology, with a radioactive labelling of the sample, and with cDNA probes spotted on the membrane. We do first an hybridisation with a vector oligonucleotide, that match all spot on the chip. This is to adjust for the different quantities spotted on the chip. Then we hybridize with the complex set issued by RTPCR from the sample. I can't figure out really what it implies when i use the different methods in Bioc. As it is a single channel procedure i'd think it's closer to affy chips for example. But in the other hand, as we do the first hybridization, and as it is a cDNA technology, i wrote a read function to load my data in a RGlist object, and then use it in limma. In this function, R channel correspond to the complex set, and G channel to the vector set. (two successive hybridisations on the same membrane). But in fact my ratio "complex / vector" has not really the same meaning as "sample labelled with Cy5 / reference sample labelled with Cy3". These questions raised as i was reading the monograph on bioconductor. The methods used to identify differentially expressed genes are not the same for an affybatch, and for a batch using limma for example. As i'm a wet-lab biologist, and as my skills in statistics are low, i don't really understand if it's the same or not, but i would say it's not ? So did you use different approaches because in one case your are using a signal from a single channel, oligo chip, and on the other one, a dual channel with a ratio of expression of a cDNA chip ? If it is true, the fact that my ratio "doesn't really represent the same thing as a dual channel cDNA chip" would mean that i have to use a specific method, and then which is the best. I don't know if what i say is clear, but i would appreciate your comments on this. Do you think in this case i should consider my ratio as a normalised (a first step of) value, and then think as a single channel ? Thanks for any help, Julien

Hi Julien, in the past I analyzed a lot of nylon membrane data. For normalization I used "vsn" (which can be applied to single-channel data), and for finding differentially expressed genes, methods in "limma" or "genefilter" are appropriate, depending on your experimental design (which you didn't describe). I never found any use for the "vector" hybridizations, and just treated the data as from any other single-channel method. If you use arrays from one PCR- and spotting batch, the probe effect (different quantities spotted on the chip) should be the same for all data points from one probe and hence cancels out for differential expression calculations. However, the "if" from above, as well as the problems from stripping and re-using the filters, meant that the data quality was much worse than with two-color glass slides or commercial oligonucleotide arrays. These may seem more expensive on first sight, but when you look at value for money the reverse appears to be true. Cheers Wolfgang TEXTORIS Julien wrote: > Hi all, > > I'm using a nylon technology, with a radioactive labelling of the > sample, and with cDNA probes spotted on the membrane. > > We do first an hybridisation with a vector oligonucleotide, that match > all spot on the chip. This is to adjust for the different quantities > spotted on the chip. > Then we hybridize with the complex set issued by RTPCR from the sample. > > I can't figure out really what it implies when i use the different > methods in Bioc. As it is a single channel procedure i'd think it's > closer to affy chips for example. But in the other hand, as we do the > first hybridization, and as it is a cDNA technology, i wrote a read > function to load my data in a RGlist object, and then use it in limma. > In this function, R channel correspond to the complex set, and G channel > to the vector set. (two successive hybridisations on the same membrane). > > But in fact my ratio "complex / vector" has not really the same meaning > as "sample labelled with Cy5 / reference sample labelled with Cy3". > > These questions raised as i was reading the monograph on bioconductor. > The methods used to identify differentially expressed genes are not the > same for an affybatch, and for a batch using limma for example. As i'm a > wet-lab biologist, and as my skills in statistics are low, i don't > really understand if it's the same or not, but i would say it's not ? > So did you use different approaches because in one case your are using > a signal from a single channel, oligo chip, and on the other one, a dual > channel with a ratio of expression of a cDNA chip ? > > If it is true, the fact that my ratio "doesn't really represent the same > thing as a dual channel cDNA chip" would mean that i have to use a > specific method, and then which is the best. > > > I don't know if what i say is clear, but i would appreciate your > comments on this. Do you think in this case i should consider my ratio > as a normalised (a first step of) value, and then think as a single > channel ? > > Thanks for any help, > > Julien > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor -- Best regards Wolfgang ------------------------------------- Wolfgang Huber European Bioinformatics Institute European Molecular Biology Laboratory Cambridge CB10 1SD England Phone: +44 1223 494642 Fax: +44 1223 494486 Http: www.ebi.ac.uk/huber