Dear Bugs users,
I am currently struggling with the approach of Mike West at al
of classification of microarray experiments according to a
regression on the scores of a singular value decomposition
of the original gene expression profiles.
http://citeseer.nj.nec.com/west00dna.html
The model formulation is as follows:
model {
for (i in 1:nCHIPS) {
tau[i] ~ dgamma(1.5, 1.5)
wtau[i] <- tau[i]*d[i]*d[i]
gamma[i] ~ dnorm(0, wtau[i])
eta[i] <- exp(inprod(gamma[],w[i,] ))
pi[i] <- (eta[i]/(1+eta[i]))
z[i] ~ dbin(pi[i], 1)
}
}
Unlike the original proposition which does probit, I am following
a logistic regression approach here (pi is expit of X times beta).
In my attempt on my own data, the means of the posterior densities
appear quite sensible, especially when compared to visual analysis.
Unfortunately, the chain diverges and I get a Trap message "undefined
real result". This happens sometimes instantly and sometimes after
a couple of 100 iterations, depending on initial values. As an
additional nuisance, the posterior densities appear to be rather
kurtotic, with the 2,5 and 97,5 percentiles beyond the mean+-2stdev
range.
Has anyone used this approach with BUGS, and what is your experience
with different priors?
Greetings
Johannes Hüsing
-------------------------------------------------------------------
This list is for discussion of modelling issues and the BUGS software.
For help with crashes and error messages, first mail [log in to unmask]
To mail the BUGS list, mail to [log in to unmask]
Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html
Please do not mail attachments to the list.
To leave the BUGS list, send LEAVE BUGS to [log in to unmask]
If this fails, mail [log in to unmask], NOT the whole list
|