Figure 1. General pipeline followed in this work with description of the steps followed in the initial predictive biocomputational analysis.

Data integration of protein-protein interaction (PPI) predictions from domains co-occurrence data (CODA), gene expression similarity (GECO), interaction protein homology relationships (hiPPI) and other PPI datasets into one single PPI network model. This human interactome model is used to implement context information analyses by computing kernel distance scores between target proteins and a seed set of 116 known angiogenic proteins. These kernel scores are used to rank target proteins in a priorization list that is validated using a leave-one-out cross-validation method represented by the ROC validation curve. Finally, a set of target proteins are selected from the top ranked list of predictions for their further experimental validation.