| A new quantitative trait linkage mapping method developed in this study allows the parallel mapping of all genes of an organism for thousands of gene expression “traits”. The combination of genomic mismatch scanning (GMS) and global gene expression profiling provides the basis for this type of trait mapping, which uses gene expression profiles as molecular phenotypes. However, one difficulty associated with such study is that it deals with an enormous quantity of data with significant amount of noise. In addition, multiple testing problem becomes a serious concern when thousands of markers are mapped for thousands of gene expression “trait”. Therefore, a comprehensive statistical framework is required for building accurate linkage maps using the above promising biotechnology. To this end, an automated procedure was developed, which includes: (1) normalizing data, (2) imputing missing data, (3) identifying genes whose variation in gene expression are heritable, (4) clustering heritable genes with similar expression patterns and (5) applying genome-wide linkage mapping to these clusters. This protocol has been applied to map the genetic origins of the heritable molecular phenotype from a cross of two unrelated S. cerevisiae strains. In this study, many previously unknown clusters have been identified and linked to some regions on the yeast genome. Further characterization of these clusters may lead to significant biological findings. We also found twelve clusters that are linked to HO locus on chromosome 4 (p = 0.0001) and four clusters that are linked to both HO locus on chromosome 4 and MDS allele of CST13 on chromosome 2 (p = 0.0002). The consistency of locating the same regulatory loci for both visible macro phenotypes and invisible molecular phenotypes strongly suggests that molecular phenotype does highly correlates with macro phenotype. |
