[BioC] Using limma with contrast matrix ,replicate spots,

[Gordon K Smyth]

> Date: Mon, 24 Jan 2005 18:30:13 -0500
> From: Pita <pwilkinson_m at xbioinformatics.org>
> Subject: [BioC] Using limma with contrast matrix ,replicate spots,
> To: bioconductor <bioconductor at stat.math.ethz.ch>
>
> What is the significance of this error? What does this tell me about these
> two samples?. I added the coeficients for the donor then re-calculated the fit.
>
>
>  > donor.fit = lmFit(MA, design.III)
> Coefficients not estimable: HISS0039 HISS0050

This is not an error.  It means just what it says: the two coefficients listed cannot be
estimated.  This is because the model is over-parametrized, i.e., the columns of the design matrix
are not linearly independent.  The model will still be fit correctly but those two coefficients
will be set to NA.

Gordon