I would like to look at gene-gene correlations within a multi- factorial, mixed effects experiment. Here are the factors, with levels: Animals: 12 (6 in each strain, see below) note the animals are numbered 1:12, but I have tried 1:6 for each strain and it did not address my problem. I still wonder, however, how they should be numbered or if it makes a difference. Tissues: 3 Animal Strain: 2 I thus have 6*3*2 = 36 samples, from a total of 12 animals. Tissue and strain are fixed effects and, for other purposes, of some interest, the animals are random effects and not of interest. All I want to look at now is the correlation of the continuous variable of expression between 2 genes. In biological terms, I want to look for evidence of co-regulation. It looks to me like I should use lmer/lmer2 for this, but I have not set up a model with this function before. I tried to follow an example in "The R Book" by Crawley, pp. 648-50, without success (see my code chunk below). mod <- lmer(gene.mat[2,] ~ + gene.mat[1,] + (1|Strain/Tissue/Rat)) I strongly suspect my error is in the model, but in case it might help, below is the actual output. Note my model terms above have been decrypted. I did check to make sure that the phenoData columns are correct, and they are. Thanks, Mark output: Error in pData(exprSet)$Treatment:factor : NA/NaN argument In addition: Warning messages: 1: In pData(exprSet)$Treatment:factor : numerical expression has 36 elements: only the first used 2: In inherits(x, "factor") : NAs introduced by coercion Enter a frame number, or 0 to exit 1: lmer(gene.mat[2, ] ~ +gene.mat[1, ] + (1 | factor(pData(exprSet)$Treatment) 2: lmerFactorList(formula, mf, fltype) 3: lapply(bars, function(x) eval(substitute(as.factor(fac)[, drop = TRUE], lis 4: FUN(X[[1]], ...) 5: eval(substitute(as.factor(fac)[, drop = TRUE], list(fac = x[[3]])), mf) 6: eval(expr, envir, enclos) 7: as.factor(factor(pData(exprSet)$Rat):(factor(pData(exprSet)$Tissue):(p Data( 8: is.factor(x) 9: inherits(x, "factor") > sessionInfo() R version 2.7.0 Under development (unstable) (2008-02-17 r44506) x86_64-unknown-linux-gnu locale: LC_CTYPE=en_US.UTF-8;LC_NUMERIC=C;LC_TIME=en_US.UTF-8;LC_COLLATE=en_US .UTF-8;LC_MONETARY=en_US.UTF-8;LC_MESSAGES=en_US.UTF-8;LC_PAPER=en_US. UTF-8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=en_US.UTF-8 ;LC_IDENTIFICATION=C attached base packages: [1] tools stats graphics grDevices datasets utils methods [8] base other attached packages: [1] lme4_0.99875-9 Matrix_0.999375-4 lattice_0.17-6 [4] affy_1.17.5 preprocessCore_1.1.5 affyio_1.7.12 [7] Biobase_1.17.13 loaded via a namespace (and not attached): [1] grid_2.7.0 tcltk_2.7.0 > -- Mark W. Kimpel MD ** Neuroinformatics ** Dept. of Psychiatry Indiana University School of Medicine 15032 Hunter Court, Westfield, IN 46074 (317) 490-5129 Work, & Mobile & VoiceMail (317) 204-4202 Home (no voice mail please) mwkimpel<at>gmail<dot>com

Having not gotten a reply and having found a list dedicated to just these types of questions on mixed models, I am going to repost on R-SIG-ME. Mark Mark W. Kimpel MD ** Neuroinformatics ** Dept. of Psychiatry Indiana University School of Medicine 15032 Hunter Court, Westfield, IN 46074 (317) 490-5129 Work, & Mobile & VoiceMail (317) 204-4202 Home (no voice mail please) mwkimpel<at>gmail<dot>com ****************************************************************** Mark W Kimpel wrote: > I would like to look at gene-gene correlations within a multi- factorial, > mixed effects experiment. Here are the factors, with levels: > > Animals: 12 (6 in each strain, see below) note the animals are numbered > 1:12, but I have tried 1:6 for each strain and it did not address my > problem. I still wonder, however, how they should be numbered or if it > makes a difference. > > Tissues: 3 > Animal Strain: 2 > > I thus have 6*3*2 = 36 samples, from a total of 12 animals. > > Tissue and strain are fixed effects and, for other purposes, of some > interest, the animals are random effects and not of interest. All I want > to look at now is the correlation of the continuous variable of > expression between 2 genes. In biological terms, I want to look for > evidence of co-regulation. > > It looks to me like I should use lmer/lmer2 for this, but I have not set > up a model with this function before. I tried to follow an example in > "The R Book" by Crawley, pp. 648-50, without success (see my code chunk > below). > > mod <- lmer(gene.mat[2,] ~ + gene.mat[1,] + (1|Strain/Tissue/Rat)) > > I strongly suspect my error is in the model, but in case it might help, > below is the actual output. Note my model terms above have been > decrypted. I did check to make sure that the phenoData columns are > correct, and they are. > > Thanks, Mark > > output: > Error in pData(exprSet)$Treatment:factor : NA/NaN argument > In addition: Warning messages: > 1: In pData(exprSet)$Treatment:factor : > numerical expression has 36 elements: only the first used > 2: In inherits(x, "factor") : NAs introduced by coercion > > Enter a frame number, or 0 to exit > > 1: lmer(gene.mat[2, ] ~ +gene.mat[1, ] + (1 | > factor(pData(exprSet)$Treatment) > 2: lmerFactorList(formula, mf, fltype) > 3: lapply(bars, function(x) eval(substitute(as.factor(fac)[, drop = > TRUE], lis > 4: FUN(X[[1]], ...) > 5: eval(substitute(as.factor(fac)[, drop = TRUE], list(fac = x[[3]])), mf) > 6: eval(expr, envir, enclos) > 7: > as.factor(factor(pData(exprSet)$Rat):(factor(pData(exprSet)$Tissue): (pData( > 8: is.factor(x) > 9: inherits(x, "factor") > > > sessionInfo() > R version 2.7.0 Under development (unstable) (2008-02-17 r44506) > x86_64-unknown-linux-gnu > > locale: > LC_CTYPE=en_US.UTF-8;LC_NUMERIC=C;LC_TIME=en_US.UTF-8;LC_COLLATE=en_ US.UTF-8;LC_MONETARY=en_US.UTF-8;LC_MESSAGES=en_US.UTF-8;LC_PAPER=en_U S.UTF-8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=en_US.UTF -8;LC_IDENTIFICATION=C > > attached base packages: > [1] tools stats graphics grDevices datasets utils methods > [8] base > > other attached packages: > [1] lme4_0.99875-9 Matrix_0.999375-4 lattice_0.17-6 > [4] affy_1.17.5 preprocessCore_1.1.5 affyio_1.7.12 > [7] Biobase_1.17.13 > > loaded via a namespace (and not attached): > [1] grid_2.7.0 tcltk_2.7.0 > > > > -- > > Mark W. Kimpel MD ** Neuroinformatics ** Dept. of Psychiatry > Indiana University School of Medicine > > 15032 Hunter Court, Westfield, IN 46074 > > (317) 490-5129 Work, & Mobile & VoiceMail > (317) 204-4202 Home (no voice mail please) > > mwkimpel<at>gmail<dot>com > > ****************************************************************** >