[R] Several questions about MCMClogit

[j t]

Hello everybody,

I'm new to MCMClogit.  I'm trying to use MCMClogit to fit a logistic
regression model but I got some warnings I can't understand.

My input data X is 32(tissue sample)*20(genes) matrix, each element in
this matrix corresponds to the expression value of one particular gene
in one of 32 samples. And the Y presents the corresponding classes
(0-non cancer, 1-cancer) for those 32 samples. The formula is Y~. All
other parameters are default.

This is the output message:
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
The Metropolis acceptance rate for beta was 0.00000
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
Warning messages:
1: In glm.fit(x = X, y = Y, weights = weights, start = start, etastart
= etastart,  :
  fitted probabilities numerically 0 or 1 occurred
2: In glm.fit(x = X, y = Y, weights = weights, start = start, etastart
= etastart,  :
  fitted probabilities numerically 0 or 1 occurred

Question 1:
What are the meanings of two warning messages?

Question 2:
How could I use the regression coefficients to predict for other data,
in other words, how could I extract those regression coefficients from
the result of MCMClogit.

I know maybe my questions are some basic but it already bothered me
for several days. I hope somebody can give me some hint about them.
BTW, is there any specific manual or examples for using MCMClogit, the
manual in package is not enough to figure out my problem.

Thanks and regards,

Chao