Welcome to MARIMBA!
Welcome to MARIMBA, the Molecular Annotation Resource for Integrating Microarray with Bayesian Analysis. MARIMBA is under development at the University of Michigan.
High throughput microarray technology targets the interactions and functions of thousands of genes in a multitude of model organisms. One challenge in microarray data analysis is determining the interactions of these genes with varying levels of prior knowledge. A recent advent in the prediction of gene interaction networks, such as in bacterial pathogenesis and host defense, is the implementation of Bayesian techniques towards microarray analysis. Several Bayesian methods have been used to generate new hypothetical causal networks from microarray data. Despite an abundance of Bayesian tools in the literature and web, implementing Bayesian analysis towards microarray analysis is meticulous and sometimes problematic.
MARIMBA is designed to simplify Bayesian analysis as applied to microarray data. We offer several tools to aid in your analysis. First, MARIMBA uses the powerful BANJO system developed at Duke University for static and dynamic Bayesian analysis, and converts this system to a web-accessible microarray analysis tool. MARIMBA provides options for users to either deposit their microarray data into our system, or to select from any sample (GSM) and/or data (GDS) available at the NCBI GEO website. MARIMBA automates microarray data selection and modeling, and displays the Bayesian network results. Key features of the MARIMBA pipeline include filtering and clustering options, a probe conversion/annotation tool, a BANJO parameter selection interface, and an interactive network output by MARIMBA. MARIMBA is implemented using a PHP interface and MySQL database.
- Static and Dynamic Bayesian Analysis
- Implementation of BANJO system created at Duke University
- Filtering Algorithms
- Clustering Algorithms
- K-Means (Pycluster)
- SOM (*coming soon*)
- Probe Conversion Tool
- Currently available for a select number of chips