[BioC] limma ebayes

[Gordon Smyth]

Ken,

As Naomi says, you have too many parameters for your data. You need to 
check your inputs to see if you're actually giving the function any data at 
all. Try

summary(fit$coef)
summary(fit$df.residual)
summary(fit$sigma)

The error message is generated by the fact that all your fit$df.residual 
entries are zero! This may be caused by a error earlier meaning that you're 
not passing the data that you think you are.

Gordon

> >>> Naomi Altman <naomi at stat.psu.edu> 01/16/04 3:06 PM >>>
>Dear Ken,
>This error message is due to having too many parameters for the number
>of
>observations.  You need to provide:
>
>the experiment design (i.e. how many arrays for each condition and the
>set
>of conditions)
>the design matrix you sent to the program
>
>Without this information, no-one can assist you.
>
>--Naomi Altman
>
>
>At 10:11 AM 1/16/2004, kninh at mdanderson.org wrote:
> >Hello Limma Developers,
> >I got this error when calling ebayes after  lm.series:
> >
> >Error in ebayes(fit) : No residual degrees of freedom in linear model
>fits
> >
> >I tried both limma 1.3.1 and 1.3.12.   I appreciate your help.
> >
> >thanks,
> >Ken Ninh