[BioC] Fisher Information Matrix in edgeR

[Mark Robinson]

Dear Tieming,

Actually, approx.expected.info() is a very old function (only used sparingly in early versions), was only available for the dispersion parameter conditional likelihood (not the regression parameters) and, as you have found, is not valid for GLMs.

I don't think (and Gordon can correct me if I'm wrong) that the current software has an easy way to access the observed or expected Fisher information.  But, they shouldn't be too hard to compute, e.g. bottom of page 14:

http://www.stat.umn.edu/geyer/5931/mle/glm.pdf

Best, Mark


----------
Prof. Dr. Mark Robinson
Bioinformatics
Institute of Molecular Life Sciences
University of Zurich
Winterthurerstrasse 190
8057 Zurich
Switzerland

v: +41 44 635 4848
f: +41 44 635 6898
e: mark.robinson at imls.uzh.ch
o: Y11-J-16
w: http://tiny.cc/mrobin

----------
http://www.fgcz.ch/Bioconductor2012



On 09.10.2012, at 02:36, Ji, Tieming wrote:

> Dear Bioconductor Authors,
> 
> I fitted a generalized linear models using glmFit() in edgeR. I wanted to extract the estimated fisher information matrix (or the estimated inverse fisher information matrix) from the fitting. The function approx. expected.info() in edgeR seems to be the right function to use. However, I have two problems when using it: (1) It does not work with the GLM since it requires the pseudo.alt argument which is provided only in the one-way layout framework; (2) I have ~30,000 tags, and it takes more than an hour to run. Could you please let me know how to extract the estimated (inverse) fisher information matrix?
> 
> Thanks,
> Tieming
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor