A flexible pipeline for experimental design, processing, and analysis of microarray data

Abstract

We created a web-based microarray data analysis pipeline for managing the volumes of data created by production microarray experiments. Experiments are formalized by grouping array data into hierarchies based on types such as 'dye swap' or 'replicate'. Grouping determines the analysis to be performed and enables the tool to automatically generate reports and charts appropriate to the experiment results. Subsets of data across arrays may also be hierarchically grouped into types such as 'gene' or 'list'. The group hierarchy is similar to a document object model (DOM), which enables queries to be posed in an XPath or XQuery language. Analyzer modules provide the complicated statistical processing and may be custom written or implemented as wrappers around existing tools. For speculative data analysis or publication, the results may be exported to a standard format.