[BioC] question about the error from function intraspotCorrelation

[Gordon Smyth]

At 04:50 AM 31/07/2004, Ren Na wrote:
>I tried to do individual channel analysis of two-colour microarray data.
>I sort of followed the example in function lmscFit. When I ran to the
>the following step, I got the error,
>corfit <- intraspotCorrelation(MA, design)
>Error in if (dev < devold - 1e-15) break :
>         missing value where TRUE/FALSE needed
>
>I checked M and A, there is no missing value. Does anyone know what
>cause this error?

The error message is from the function remlscore() in the statmod package, 
and indicates a failure of the iterative algorithm used to fit a REML 
model. remlscore() is usually very reliable. One possible explanation is 
that your data is degenerate in some way so that a model can't be sensibly 
fit. If you provide a reproducible example of the error, I will have a look 
at it.

Gordon

>Thank you very much.