Network analysis of reverse phase protein expression data: characterizing protein signatures in acute myeloid leukemia cytogenetic categories t(8;21) and inv(16).

Acute myeloid leukemia (AML) patients present with cancerous cells originating from bone marrow. Proteomic data on AML patient cells provides critical information on the key molecules associated with the disease. Here, we introduce a new computational approach to identify complex patterns in protein signaling from reverse phase protein array data. We analyzed the expression of 203 proteins in cells taken from AML patients. Dominant overlapping protein networks between subtypes of AML patients were characterized computationally, through a paired t-test approach looking at relative protein expression. In the first application of this method, we compared recurrent cytogenetic abnormalities inv(16) and t(8;21), both affecting core-binding factor (CBFβ), to normal CD34(+) cells and to each other. Six hundred seventy-eight sets of proteins were identified as significantly different in both inv(16) and t(8;21) compared to controls, at the Bonferroni number, α < 2.44 × 10(-6) . We strengthened our predictions by comparing results to those obtained using lasso regression analysis. Signaling networks were constructed from the protein pairs that were significantly different in the t-test and lasso regression analysis. Predicted networks were also compared to known networks from public protein-protein interaction and signaling databases. By characterizing unique "protein signatures" through this rapid computational analysis, and placing them in the context of canonical biological networks, we identify signaling pathways distinct to subcategories of AML patients.