[BioC] What do the results mean?

[Naomi Altman]

We seem to get quite a few questions of the following type:


>Are they the genes up/down regulated due to sample A or sample B?

As a statistician, I want to urge you to LOOK at your data before (or at 
least along with) your p-values.  For at least a few of your significant 
and nonsignificant genes, you should look at the normalized data and the 
treatment means and make sure that the results are sensible.  I have been a 
statistician for over 25 years, and I still do this with EVERY analysis I 
do. Firstly, it helps you decide if there is something wrong with the data 
- e.g. big outliers that are affecting the analysis, or missing values that 
really should be there, or the wrong column of the data 
selected.  Secondly, it helps you understand features of the analysis, such 
as whether the genes have been up or down regulated, or why certain 
interactions are important.  Finally, (and perhaps I am less careful than 
others), errors often creep into my analysis steps and these are usually 
obvious if I look at the data along with the results, rather than just the 
results.  (For example, I once failed to notice that an investigator had 
used 9999 as a missing value code - we had some highly significant results 
due to including this as valid data.)


>_______________________________________________
>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