[BioC] Re: Bioconductor Digest, Vol 6, Issue 33

[Leslie Cope]

> 
> The controls that Affy uses are spiked-in transcripts that are designed
> to indicate problems in the hybridization step, which would be
> indicative of a 'bad' chip.
> 
> The RNA digestion plot gives a qualitative idea of two things; the
> amount of RNA degradation that occurred during the preparation of your
> RNA (due presumably to RNAases), and how well the second strand
> synthesis went in the preparation of your samples.
> 
> Since these are two completely different things, you cannot compare the
> results and ask which one you should believe. You can believe both and
> there is no contradiction.
> 
> As an aside, I almost always see a 'significant' slope to the curves in
> the RNA digestion plot. I don't necessarily think this is a critical
> problem. However, if the curves have completely different slopes, I
> start to get worried (not that there is much I can do at that late
> stage...).
> 
> Jim
> 
An addition.  

I would ignore the p-values from the RNA degradation plots.  They are
not remotely accurate, and will either disappear entirely in future
versions of the function, or be calculated very differently.  The slope
has a much lower coefficient of variation than the best of the
individual housekeeping genes.  Once experience has shown us which slope
values are cause for concern in a given chip type, the slope may be a
more sensitive indicator of problems than housekeeping genes.  That
experience will be essential though for correct interpretation.

Leslie Cope