Fangxin, Yes it looks like the data is less than ideal. The last chip certainly looks dubious and should probably be repeated. I would definetely check with the experimenter how the samples were harvested and process and CONFIRM that they are true biological replicates. It's amazing how many lab plant biologists see pooled samples from a bulk of plants grown at the same time as biological replicates when they are clearly not. Looking at the RNA-deg plot (and sample labels) I guess they could be epidermis cells or cell layer versus bulk stem or the underlying stem tissue. If the tissue preparation required for the different sample types was significantly different then this is the most likely reason for the similarities seen in the RNA-deg plot, eg: it would take much longer to take epidermal peels then a stem section and so RNA degradation could be higher. Or the extracts could have a different composition and something (eg:sugars) may effect the RNA extraction efficiency or quality. Alternatively labelling or hyb in different batches could also lead to the same effect. I find hist, RNA deg, AffyPLM and a simple RMA norm followed by plot(as.data.frame(exprs(eset.rma))) can answer in most cases for why it didn't work, or won't work - in the rare case when someone asks for QC before rather than after they realise the data is strange ;-) Cheers, Matt >>>>>>>>>>> previous >>>>>>>>>> Hi Naomi, Thank you for your help. > It looks to me as if there is a problem in this experiment. I cannot > speak for the efficacy of the RNA degradation plot. But unless a > large amount of differential expression occurs in this experiment, > the very close similarity between the duplicates compared to the > other conditions leads me to thing that these duplicates were either > not biological replicates, or the duplicates were processed together > causing correlation. I know that the replicates are biological replicates, so I think very likely that they processed the duplicates together ( will check with the experimenter) However, if this is the case, what we can do to? It violates the assumption of Limma ? Maybe Rank product can be a solution since it computes 4 ratios among duplicates from two conditions? Many thanks! fangxin > I have seen this type of thing with spotted arrays when arrays > processed in a single batch are much more similar than biological > replicates processed on different days. > > --Naomi > > At 08:12 AM 1/18/2006, James W. MacDonald wrote: >>fhong at salk.edu wrote: >> > Dear list, >> > >> > I have this 8 affy arrays under 2*2 factorial design, with duplicates >> > under each condition. The RNA degradation plot worries me since the >> slopes >> > from 8 arrays are so different, with duplicates under each condition >> as >> > one group (see the QC plots at http://cactus.salk.edu/temp/QC-1.jpeg) >> > I would suspect that these arrays were processed under different >> levels >> > if amplification. >> > >> > My problem is how to handle this data set beside doing the >> normalization? >> > Will this pattern seriously bias the result? I read some previous >> message >> > about this topic, just hope to get more information. >> >>I find that the RNA degradation plots are less useful for indicating >>possible problems than the density plots. If the density plots are all >>reasonably similar, in my experience the normalization should be fine. >>Another excellent plot for detecting problems is the residual plot in >>the affyPLM package. >> >> >>Best, >> >>Jim >> >> >> > >> > Many thanks! >> > Fangxin

Hi Matthew, Thank you very much for your help. > It's amazing how many > lab plant biologists see pooled samples from a bulk of plants grown at > the same time as biological replicates when they are clearly not. I would think that all plants under experiment shoudl be grown at the same time without different conditions/treatments. Biological replicates should be tissue samples from differnt groupd of plants, say sample from 50 plants as replicate1 and sample from another 50 as replicate 2. Do you think that biological replicates should be grown at different time? > I find hist, RNA deg, AffyPLM and a simple RMA norm followed by > plot(as.data.frame(exprs(eset.rma))) can answer in most cases for why it > didn't work, or won't work - in the rare case when someone asks for QC > before rather than after they realise the data is strange ;-) This actually pull out another question: when % of differential genes is large, which normalization better works better? http://cactus.salk.edu/temp/QC_t.doc Take a look at the last plot, which clearly indicate homogeneous within replicates and heterogeneous among samples. (1) Will stem top and stem base differ so much? Or it is the preparation process bring in extra correlaton within replicates. (2) when % of differential genes is large, which normalization better works better? many thanks! Fangxin -------------------- Fangxin Hong Ph.D. Plant Biology Laboratory The Salk Institute 10010 N. Torrey Pines Rd. La Jolla, CA 92037 E-mail: fhong at salk.edu (Phone): 858-453-4100 ext 1105