PATH-22. BRAINGENEEXPRESS: A WEB APPLICATION FOR DATA VISUALIZATION OF BRAIN TUMOR RESEARCH

During the past decade, several public databases have provided researchers with an unparalleled amount of data on brain tumors, like The Cancer Genome Atlas, Gene Expression Omnibus, and the recent MSK-IMPACT project. How to comprehensively integrate and easily visualize these data becomes crucial for researchers to extract any meaningful information for the benefit of translational research. We aim to build a web application, BrainGeneExpress, for visualization of data relevant to glioblastoma study to help researchers rapidly access the results by selecting genes of interest, available databases, and appropriate methods. Our web application will consolidate a large amount of data analysis results from both clinical and bioinformatics perspectives, which are not available in already existing web portals. We used Shinny by Rstudio to build the web application. It currently includes: 1) Survival analysis on overall survival and time-to-recurrence using Cox proportional hazards model with and without restricted cubic spline with adjustment of user-selected clinical factors. Different variable selection and missing data imputation methods were also available; 2) Correlation analysis: including correlation between two gene sets using canonical correlation analysis; correlation between one gene and a selected array of genes or the entire gene pool, with significant correlated genes filtered by different criteria, and visualized in both table and heatmap, and clustering analysis was also applied; 3) Gene set enrichment analysis; 4) Recursive partitioning analysis. All results can be easily obtained, and downloaded into different formats, with table notes and figure legends included. Successful completion of this web application will trigger the exploration of more brain tumor types, databases, and analysis methods aimed at including lower grade gliomas, pediatric gliomas, more GSE databases, and machine learning methods. It will greatly enhance the knowledge of glioma for researchers through data mining from different perspectives and make brain tumor research more efficient.