Development of a relational database allowing filtering biologically relevant groups of genes in microarray data statistical analysis.

The objective of this project is to reduce the background noise which limits the interpretation of experiments using microarrays, by integrating information from genomic databases to statistical analysis. Statistical methods generate a high background noise, filtered today only downstream of the analysis, by analytical tools and bibliographic biological information. They are unable to filter false negatives according to statistical criteria.

These criteria are generally used independently of each other and independently for each gene. Their consideration is neither automatic nor weighted.
To filter the results of the analysis of microarrays, a promising way is to submit to the statistical analysis subsets of genes between which relations have been established in reference to bioinformatics databases. The aim of the project is to develop a relational database specifically dedicated to the analysis of these results, by crossins information from numerous available databases.
A request may be structured via an expert system and this information may be integrated at various levels of the decision-making statistical methodology to increase substantially its power.