[R-sig-ME] Pedigreemm on lme4-1.0
Ben Bolker
bbolker at gmail.com
Sat Feb 15 03:17:15 CET 2014
On 14-02-14 06:07 PM, Joehanes, Roby (NIH/NHLBI) [F] wrote:
> Hi Ben:
>
> The package could install and the examples in the manual could run.
> However, all the examples given in the pedigreemm technical paper do not:
> milk <- within(milk, sdMilk <- milk / sd(milk))
> fm1 <- pedigreemm(sdMilk ~ lact + log(dim) + (1|id) + (1|herd), data =
> milk, pedigree = list(id = pedCowsR))
>
> The error message is:
> Error in .sortCsparse(.Call(dtCMatrix_sparse_solve, a, b)) :
> Dimensions of system to be solved are inconsistent
>
> The same goes to these examples (same error message):
> milk <- within(milk, idPE <- id)
> fm2 <- pedigreemm(sdMilk ~ lact + (1|id) + log(dim) + (1|idPE) + (1|herd),
> data = milk, pedigree = list(id = pedCows))
>
> fm3 <- pedigreemm(NCM ~ birth + calvingYear + (1|sire) + (1|herd), data =
> mastitis, pedigree = list(sire = pedSires), family = "poisson")
>
> fm4 <- pedigreemm(mastitis ~ birth + calvingYear + (1|sire) + (1|herd),
> data = mastitis, pedigree = list(sire = pedSires),
>
>
> cf: http://www.journalofanimalscience.org/content/88/2/497.full.pdf+html
>
> Updating/recompiling Rcpp, RcppEigen, and lme4 do not help. I see no test
> codes at all in pedigreemm-0.3-1 package. As such, pedigreemm at the
> present state could only be used for toy examples. Or this could be a
> problem exclusively on my side.
I don't think so, I can replicate your first example.
>
> Plus, I remember that I submitted a patch to allow specification of one
> observation per ID. It seems that the patch did not transfer over. I
> always get this error message:
> Error in checkNlevels(reTrms$flist, n = n, control) :
> number of levels of each grouping factor must be < number of observations
> http://r-forge.r-project.org/tracker/index.php?func=detail&aid=1928&group_i
> d=60&atid=300
Your patch was appreciated, but superseded by fairly major
architectural changes we made. There is a fairly straightforward way now
to bypass this test:
control=lmerControl(check.nobs.vs.nlev="ignore")
This really should be done within the call in pedigreemm, but it can be
enabled globally as follows:
options(lmerControl=list(check.nobs.vs.nlev="ignore"))
> I can try to fix this problem, but I will need guidance from you and
> others since I have been absent from the lme4 development for 1.5 years
> now. The patch above is now obsolete with lme4-v1.0.
>
> Sincerely,
> Roby
>
>
>
>
>
> On 2/14/14 5:29 PM, "Ben Bolker" <bbolker at gmail.com> wrote:
>
>> On 14-02-14 12:22 PM, Joehanes, Roby (NIH/NHLBI) [F] wrote:
>>> Hi all:
>>>
>>> Just wondering if there is any people working on pedigreemm so that
>>> it can run on the new lme4 version 1.0? In current version of lme4
>>> 1.0, pedigreemm is broken. May I know the latest status on
>>> pedigreemm?
>>>
>>> Thanks, Roby
>>>
>>> _______________________________________________
>>> R-sig-mixed-models at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>>
>>
>> I'm a little surprised, because the current CRAN page suggests it's
>> OK: http://cran.r-project.org/web/packages/pedigreemm/index.html -- this
>> page wouldn't exist, I think, if the package didn't at least pass its
>> checks with the new version of lme4 (and it says that it requires lme4
>>> = 1.0).
>> Perhaps you could provide a little more information for those of us
>> who don't use pedigreemm on a regular basis -- does it fail to install?
>> Installs but doesn't pass checks/examples fail? Examples run but give
>> incorrect answers? Some particular cases don't work?
>> Posting here seems perfectly reasonable, but I would also suggest
>> contacting the maintainer (listed at the URL above) as well. If that
>> maintainer is unwilling or unable to fix whatever's broken, then
>> presumably someone else will have to step forward to act as maintainer ...
>>
>> I'm willing to help fix what's broken, in particular on the
>> lme4-interface end, but would need to get *specific* examples of what's
>> not working ...
>>
>> Ben Bolker
>
More information about the R-sig-mixed-models
mailing list