PIPs: Human Protein-Protein Interaction Prediction

To query a protein of interest enter the IPI, RefSeq or UniProt reference in the Quick Search box (left) and click Submit. For further methods of searching the database click on the Advanced Search link.

The predictions stored in the PIPs database are made by a naïve Bayesian classifier which calculates the likelihood ratio of interaction between two proteins given the presence of features that include: gene co-expression, orthology to known interacting protein pairs, domain co-occurrence, co-localisation of proteins within the cell and the co-occurrence of post translational modifications. The final likelihood ratio takes account of the local topology of the predicted interaction network and is based on the premise that if two proteins A and B share interactions with other proteins, there is a higher likelihood that A and B will interact. For full details on the method, please see Scott and Barton (2007).

Please refer to the help document for further details about how to use this site.