Bridging Microarray Platforms To Extend the Utility of Gene Expression Profiles

Gene expression profiling experiments provide a wealth of data for use in the analysis of the transcriptome and have yielded results that have helped in the classification of disease type and suggestions of possible biological pathways. Two major types of platforms exist, cDNA platforms where ESTs are spotted onto a glass microscope slide and oligonucleotide platforms where each gene is represented by small lengths of approximately 20 base pairs of DNA. Due to this difference in set-up and subsequent use the two platforms tend to yield different results when the data is analysed by a common method such as a clustering algorithm. This project seeks to find a general tool to cluster expression profile results from both sets of data simultaneously. A new clustering algorithm that has been presented at the Royal Statistical Society London in May 2004 will be utilised in the tool. Its claimed advantage over existing algorithms is that it can cluster objects on only small subsets of their attributes that show a signal above or below the background noise of the majority of the data. This type of data structure is typical of that produced by the microarray platforms and suggests the application of the new algorithm to their study.