[BioC] problems with MANOR

J.delasHeras at ed.ac.uk J.delasHeras at ed.ac.uk
Thu May 10 19:34:37 CEST 2007

Quoting James Morris <slimjimbob1 at hotmail.com>:

> Hello everyone,
> I am trying to use the MANOR spatial normalisation package but I   
> have got a problem
> I get this
> message when i calculate the log2 ratios
> Warning message:
> NaNs produced in: log(x, base)
> and when I try to normalise I get
> Warning message:
> NaNs produced in: log(x, base)
> any help would be most appreciated
> James
> the script I am using:

Hi James,

I think the culprit is here:

## compute a log-ratio
ac$arrayValues$Ratio[ac$arrayValues$Ratio==Inf] <- NA
ac$arrayValues$LogRatio <- log(ac$arrayValues$Ratio,2)

It's not unusual to have local background estimates that are larger  
than the actual signal, so you get a negative signal after background  
substraction... and the error you get is produced when trying to  
calculate the log of a negative number.

You will have to either ignore the warnings (but you will lose the  
spots that produce the error), or use a different background estimate,  
or use a different background correction method, or even not do  
background correction at all (if the slides are decent, no background  
correction is often better than background substraction).

I use the 'limma' package, which offers a nice range of methods for  
both background correction and normalisation. In particular, the  
'normexp' background correction method seems to work very well in my  
hands, when you use the correct offset. You could use limma for the  
background correction, and then continue as you are doing... I am not  
familiar with the Manor package, but I'd be surprised if you cannot  
find in limma a method to your liking.



Dr. Jose I. de las Heras                      Email: J.delasHeras at ed.ac.uk
The Wellcome Trust Centre for Cell Biology    Phone: +44 (0)131 6513374
Institute for Cell & Molecular Biology        Fax:   +44 (0)131 6507360
Swann Building, Mayfield Road
University of Edinburgh
Edinburgh EH9 3JR

More information about the Bioconductor mailing list