FYI: I did a trackback() and saw that the coefficients, sigmas, and other statistics were all produced before the sigToEnv() function caused the problem. I started a new R session, loaded the package and tried it again -- and it worked properly! I was using an older version of limma, detached it and installed/loaded the new version in the same session -- might be the problem. Thank you for your help! Yiwen -----Original Message----- From: James W. MacDonald [mailto:jmacdon@med.umich.edu] Sent: Wednesday, April 06, 2005 4:43 PM To: He, Yiwen (NIH/CIT) Cc: 'bioconductor@stat.math.ethz.ch' Subject: Re: [BioC] limma package : Error in sigToEnv(signature,fdef) : Tr ying to get slot "signature" from an object of a basic class ("NULL") wit h no slots He, Yiwen (NIH/CIT) wrote: > Thanks James, but I did do data <- as.matrix(data) before running lmFit, and > is.matrix(data) returns TRUE. ("data" is just for illustration purpose, but > thanks for pointing out the variable name issue.) > > Any other suggestions? You could always call traceback() after the error to see where things went wrong. The problem may be obvious at that point, or it may give you an idea what function to step through using debug(). An error such as yours indicates to me that lmFit() thinks you are passing some sort of an S4 object and is trying to get the contents of a slot. I don't see how this is possible with a numeric matrix (and I can't reproduce it here). Can you give a toy example that anybody could run that causes the error? > > Yiwen > -- James W. MacDonald Affymetrix and cDNA Microarray Core University of Michigan Cancer Center 1500 E. Medical Center Drive 7410 CCGC Ann Arbor MI 48109 734-647-5623