Web Analysis Database Case Study
BMC Genomics recently published a paper describing our web genomics database.
The Problem
Stanley Medical Research Institute (SMRI) provided brain tissue and clinical data for 12 different microarray studies investigating bipolar disease, schizophrenia, and depression. These studies were performed by different investigators using a range of array platforms. The investigators initially performed their own analyses of their individual studies using a range of methods, filters, and cutoffs for significance. While some commonalities were seen in the individual analyses, the studies were as notable for their seeming differences in results. SMRI was interested in performing a meta analysis of the 12 studies to help determine the real expression differences associated with the three diseases compared to normal controls.
The Solution
- Bioinformatic matching of genes/probes across studies and platforms.
- Extensive QC to identify problematic data
- Cross platform normalization of the 12 microarray studies.
- Demographic analysis to adjust for potential confounders.
- Gene specific regression models to determine regulated genes.
- Meta analysis confidence intervals for gene fold changes across studies.
- Gene detail pages summarizing all information / analysis for a given gene (~20,000 individual gene detail pages)
- Pathway / GO analysis to determine associations between pathways/GO terms and disease.
- Pathway / GO detail pages summarizing all information / analysis for a given term (~5,000 individual pathway/GO detail pages)
- Database: all of the gene annotation, sample annotation, and analysis results are stored in a MySQL database backend on the web server, and can be interactively queried via the web GUI.
- Access: via web browser
- Security: https password protected site, with multiple levels of access (e.g. more restricted access to patient clinical data fields)
