We introduce an algorithm for localizing the seizure onset zone (SOZ) of epileptic patients based on a causal influence graph estimated from ECoG recordings. In this graph the nodes correspond to the recording electrodes, while the edge weights represent the estimated pair-wise causal influence between the recorded signals. The algorithm builds upon the hypothesis that seizures start in the SOZ and then spread to surrounding areas in the brain. Thus, in the initial phases of seizures, signals recorded at electrodes close to the SOZ should have a relatively large causal influence on the rest of the recorded signals, and act as sources of information flow in the estimated graph. To find these source nodes, we propose to use a variation of the PageRank algorithm. Testing the proposed algorithm on data sets listed in the iEEG portal shows a close match between the inferred SOZ and the SOZ estimated by expert neurologists.