[BioC] Analyzing arrays with no replicates

[Naomi Altman]

I agree with Sean.  There are some ad hoc things you can do, but they 
basically amount to finding a cut-off for the fold-change and they rely 
heavily on the assumption that all genes have the same variance (which is 
unlikely).

You need a "between array" measure of variance and you do not have one.

--Naomi

At 04:02 PM 2/20/2005, Sean Davis wrote:
>>4. Once I have my filtered genes, how can I test for differential 
>>expression among the 4 groups?  I've played around with mt.maxT (with the 
>>test="f" option), but I haven't been able to get the proper class labels.
>
>You won't be able to use any standard t-test.  A t-test requires you to 
>have a standard deviation within each group.  In microarray analysis, 
>almost all analyses are done only within gene, so since you have no 
>replicates, you cannot do a t-test.  In fact, although limma would (I 
>think) allow you do do such an analysis because it uses all the genes to 
>determine a "pooled" standard deviation, I know that such an analysis will 
>not pass any kind of peer review.  If you are using this for hypothesis 
>generation, probably using fold-change is just as useful as hypothesis 
>testing.  Others may correct me on this (and I would like to hear what 
>others think), but your experimental design SERIOUSLY limits any 
>statistics you attempt.
>
>Sean
>
>_______________________________________________
>Bioconductor mailing list
>Bioconductor at stat.math.ethz.ch
>https://stat.ethz.ch/mailman/listinfo/bioconductor

Naomi S. Altman                                814-865-3791 (voice)
Associate Professor
Bioinformatics Consulting Center
Dept. of Statistics                              814-863-7114 (fax)
Penn State University                         814-865-1348 (Statistics)
University Park, PA 16802-2111