Due to the advent of high-throughput genomic technology, it has become possible to globally monitor cellular activities on a genomewide basis. With these new methods, scientists can begin to address important biological questions. One such question involves the identification of replication origins, which are regions in chromosomes where DNA replication is initiated. In addition, one hypothesis regarding replication origins is that their locations are non-random throughout the genome. In this article, we develop methods for identification of and cluster inference regarding replication origins involving genomewide expression data. We compare several nonparametric regression methods for the identification of replication origin locations. Testing the hypothesis of randomness of these locations is done using Kolmogorov-Smirnov and scan statistics. The methods are applied to data from a recent study in yeast in which candidate replication origins were profiled using cDNA microarrays.


Genetics | Microarrays | Statistical Methodology | Statistical Models | Statistical Theory

Previous Versions

Jun 15 2004 (withdrawn)
April 13, 2004