Background: Biological networks are used to represent interactions involving genes, DNA,
RNA and proteins that are able to manipulate many cellular processes.
Objective: The aim of this study is to evaluate whether prior knowledge can improve the quality of
biological networks, in particular protein-protein interaction networks and gene regulatory networks.
Method: Gene Ontology (GO) as well as three different types of semantic similarity measures were
used to assess the interaction between biological networks so as to build the corresponding filtered
networks. Both the original and the filtered networks were statistically compared against a reference
Results and Conclusion: The results confirm the effectiveness of the GO-based measure HRSS as it
improves the quality of the original network by removing many false interactions while maintaining the
true interactions. In general, the inclusion of external sources of biological information to improve the
quality of inferred knowledge (networks or any other model) is a fundamental step before the fusion of
filtered -statistically validated- intermediate results.