Exploration of non-hierarchical classification methods combined with linkage analysis to identify loci influencing clusters of co-regulated transcripts

Extensive studies have been performed to analyze variation in gene expression data by using multistage approaches, including a combination of microarray and linkage analysis. Such a method was recently used in the analysis of normal variation in gene expression by Cheung et al. (Nat. Genet. 2003, 33: 422–425) and Morley et al. (Nature 2004, 430: 743–747). Using these data, we also explored a multistage method by first performing non-hierarchical clustering for 3554 genes, which identified 114 clusters with number of genes ranging from 2 to 113. Heritabilities of the first principal component of each cluster were then estimated and 29 highly heritable clusters (i.e., h2 > 0.35) were further analyzed using variance components linkage analysis. The highest LOD score was observed on chromosome 1 (LOD = 5.36, 111.71 cM) for a cluster containing two genes [glutathione S-transferase M1 (GSTM1) and glutathione S-transferase M2 (GSTM2)] that are both located on chromosome 1p13.3. These results show the method followed in our analysis of performing cluster analysis followed by linkage analysis is another useful approach to identify chromosomal locations for genes affecting expression levels of multiple transcripts.