This is an example concerning the pathway annotation for the ER status (+/-) of breast tumors presented in the 2010 Shen and West paper. The original gene expression data include 153 tumor samples. A simple regression analysis (with tumor ER status as independent variable) using BFRM generated an estimate of the probability of association with ER status for each of the 8764 unique genes. Then using the association probability data, PROPA computed the upper and lower bounds of the marginal likelihood of association between each of the 956 pathway gene sets (originally from MSigDB C2 collection) and tumor ER status.
Step 1: Estimation of association probabilities with BFRM
Step 2: Pathway annotation with PROPA
Step 3: Presentation of pathway annotation results
All the files required for this example are archived in example_ER.zip.
by: Haige Shen, Quanli Wang & Mike West