[BioC] normexp error

[Gordon K Smyth]

> Date: Wed, 20 Jul 2005 11:27:38 +0200
> From: "Rani Elkon" <ranel at post.tau.ac.il>
> Subject: [BioC] normexp error
> To: <bioconductor at stat.math.ethz.ch>
>
> Hi,
>
> I'm trying to normalize my data using limma's normexp method. The process
> fails at some chips giving the error message pasted below.
> Does anyone know how this problem can be circumvented?
>
> (I'm using limma 2.0.2 and R 2.1.1).
>
> Thanks,
> -Rani.
>
>> dKa.RG.bgCorrect <- backgroundCorrect(dKa.RG, method="normexp", offset=50)
> Corrected array 1
> Corrected array 2
> Corrected array 3
> Corrected array 4
> Corrected array 5
> Error in optim(par = c(beta, log(sigma), log(alpha)), fn = normexp.m2loglik,
> :
>         initial value in 'vmmin' is not finite

That's a pity.  We had hoped we had made this problem very unlikely in recent versions of limma.

Try

  summary( dKa.RG$R[,6] )
  summary( dKa.RG$G[,6] )
  summary( dKa.RG$Rb[,6] )
  summary( dKa.RG$Gb[,6] )

to see if there are any problems with your data.

Other than that, I have no suggestions for a work-around at this time.

Gordon