Analysis Workflow


The SSF molecular taxonomy project seeks to perform  a comparative analysis of skin diseases leading to the determination of disease specific gene signatures that can be used in predictive modeling. We began by determining publicly available experiments that would be suitable for an integrated analysis.  The final analysis was performed on 8 microarray datasets evaluating human skin biopsies. The derived data was then normalized to bring the data into a comparable data range. Next the data was merged on shared genes resulting in 14,673 genes from 478 disease and 366 control samples. Differential expression was performed using the disease only samples comparing each disease to the rest in binary comparisons using LIMMA. The results of this analysis are available in the "Volcano Plot" module. Preliminary disease signatures were determined by taking the positive only genes from the DE lists, ranking by adjusted p-value and taking the unique top ~100 genes (some diseases had less then 100 genes). These signatures are available in the "Heat Map" module.