R help - Oct 2009 - Defining S3-methods for S4-objects: cannot coerce type 'S4' to vector of type 'integer'

In the 'doBy' package there is an esticon() function for calculating
linear contrasts for various model types. I have defined an S3-method
'esticon.mer()' for 'mer' objects from the lme4 package.
Building the package and invoking the method gives:
> esticon(fm1, c(1,1))
Confidence interval ( WALD ) level = 0.95
Error in as.integer(x) :
  cannot coerce type 'S4' to vector of type 'integer'
However, if is source the source code into R then I get what I
expected:
> esticon(fm1, c(1,1))
Confidence interval ( WALD ) level = 0.95
  beta0 Estimate Std.Error X2.value DF Pr(>|X^2|) Lower.CI Upper.CI
1     0 261.8724  6.786701 1488.888  1          0 248.5707 275.1741
The problem above arises because esticon.mer() uses the variance-covariance
matrix of the fixed effects (vcv below) in a calculation (cm is a matrix):
  cm %*% vcv %*% t(cm). The result is:

1 x 1 Matrix of class "dgeMatrix"
         [,1]
[1,] 46.05931
- and taking diag() of that matrix causes the error above... However, things
work fine if I use as.matrix() as is:  diag(as.matrix(cm %*% vcv %*% t(cm))).

- This tentatively suggests that there is no diag() method for
'dgeMatrix' objects (can hardly be the case!), but why then does my code
work when it is sourced into R???

- Does anyone have an idea about what goes wrong? )

- Is there anything 'special' that one must do in connection with
defining S3-methods for S4 objects (e.g. in connection with package building)??

I use R 2.9.2 on windows XP. I am sure that such issues must have come up for
others in the past but I've been unable to find questions/answers in the
archives. Sorry if the topic has already been covered.

Regards
Søren Højsgaard


	[[alternative HTML version deleted]]

S?ren H?jsgaard wrote:
> In the 'doBy' package there is an esticon() function for
calculating linear contrasts for various model types. I have defined an
S3-method 'esticon.mer()' for 'mer' objects from the lme4
package. Building the package and invoking the method gives:
> 
>> esticon(fm1, c(1,1))
> Confidence interval ( WALD ) level = 0.95
> Error in as.integer(x) :
>   cannot coerce type 'S4' to vector of type 'integer'
> However, if is source the source code into R then I get what I expected:
>> esticon(fm1, c(1,1))
> Confidence interval ( WALD ) level = 0.95
>   beta0 Estimate Std.Error X2.value DF Pr(>|X^2|) Lower.CI Upper.CI
> 1     0 261.8724  6.786701 1488.888  1          0 248.5707 275.1741
> The problem above arises because esticon.mer() uses the variance-covariance
matrix of the fixed effects (vcv below) in a calculation (cm is a matrix):
>   cm %*% vcv %*% t(cm). The result is:
> 
> 1 x 1 Matrix of class "dgeMatrix"
>          [,1]
> [1,] 46.05931
> - and taking diag() of that matrix causes the error above... However,
things work fine if I use as.matrix() as is:  diag(as.matrix(cm %*% vcv %*%
t(cm))).
> 
> - This tentatively suggests that there is no diag() method for
'dgeMatrix' objects (can hardly be the case!), but why then does my code
work when it is sourced into R???
> 
> - Does anyone have an idea about what goes wrong? )
A little bit of a guess, but does doBy have a NAMESPACE? If so, then
perhaps the appropriate methods are not being imported
(iimportFrom(Matrix, diag) ??)

Martin
> - Is there anything 'special' that one must do in connection with
defining S3-methods for S4 objects (e.g. in connection with package building)??
> 
> I use R 2.9.2 on windows XP. I am sure that such issues must have come up
for others in the past but I've been unable to find questions/answers in the
archives. Sorry if the topic has already been covered.
> 
> Regards
> S?ren H?jsgaard
> 
> 
> 	[[alternative HTML version deleted]]
> 
> 
> 
> ------------------------------------------------------------------------
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

-- 
Martin Morgan
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M1 B861
Phone: (206) 667-2793