Assessing Reliability of Protein Interaction Data from High-Throughput Experiments with Belief Inference

The high false positive rates associated with high-throughput experimental protein interaction data have led researchers to develop methods to assess the reliability of protein interactions generated by large-scale biological experiments.

One approach is to model the expected characteristics of true protein interaction networks, and then devise mathematical measures to assess the reliability of the candidate interactions. Saito et al. developed IG1 and IG2 to assess the reliability in topological way. However, IG1 and IG2 are local measures, which do not consider the topological properties beyond the candidate protein pair and its neighbors. We have recently developed IRAP to assess the reliability of interactions beyond the candidate protein pair with an alternative path approach. As IRAP only considers the strongest alternative path, it may overlook other interactions close to the target pair that may provide significant supporting evidence for their reliability.

We present here a new measure called ‘Interaction Reliability by Belief Inference’ (IRBI) to evaluate protein interactions by considering all the protein interactions close to the target pair. Our validation studies have shown that IRBI is an effective way for detecting reliable protein interactions.

--by Jin Chen, Wynne Hsu, Mong Li Lee, See-Kiong, APBC (poster), 2005.