High-throughput screens have begun to reveal large-scale protein interaction networks in several organisms. And knowledge mining from these networks brings significant challenges for bioinformatics, such as data quality control, topology analyses and prediction of signaling pathways. To address such challenges, several projects have been performed in Beijing Proteome Research Center.

First, we use Bayesian approaches to combine multiple heterogeneous biological evidences to assign reliability to the protein-protein interactions identified by high throughput experiments. This method shows high sensitivity and specificity to predict true interactions from the true high throughput protein-protein interaction data sets. And we’ve developed this method into an online confidence scoring system (PRINCESS). Then, based on high confidence protein interaction networks, we explore their general topological properties and find that protein interaction networks have scale-free and high-degree clustering nature as the consequence of their hierarchical organization. We also find that protein interaction networks have the small-world property with similar diameter at 4–5 and hubs play important roles in defining the network’s high degree of robustness. Not only hub proteins, we also find proteins with high SigFlux in signaling networks are also very important. Proteins with high SigFlux tend to be with high essentiality and low evolutionary rates. Because one of the main motivations of the construction of protein-protein interaction network is for signaling pathway information and it still lacks efficient computation methods, we develop an integrated strategy to mine the signaling knowledge from protein interaction network. We introduce a novel algorithm PNmerger to find the signaling subnet in protein interaction network, and then define their direction (signal flow) based on the pairwise interaction domains, which has a satisfied performance with the accuracy 89.79%, coverage 48.08% and error ratio 16.91%.

We also use these methods below to mine knowledge from the human liver protein interaction network (HLPN), which is an important part of the international Human Liver Proteome Project (HLPP). For better understanding of the functional organization of human liver proteome, a large-scale yeast two hybrid screening was used to map their interactions based on 5,026 proteins, which unbiasedly represent human liver proteome. Bioinformatics analyses and independent biochemical assays were adopted to validate the overall quality of these interactions. A network of 3,484 interactions among 2,582 proteins was established. The biological significance of this network is illustrated by the features of liver phenotype proteins, liver-specific proteins, and metabolic enzymes. Furthermore, through the investigating of the sub-network of human disease proteins and signaling pathways, dozens of potentialcandidate genes for diseases and novel regulators of specific pathways were identified. Moreover, glycerol kinase GKb was found to negatively regulate the transcription activity of NR4A1, and scaffold protein GIT2 was characterized as a novel modulator of NF-kB signaling pathway.

In summary, both these bioinformatics tools and their application in HLPN provide a reference for the establishment and analyses of large scale protein interaction network, especially those in specific cell or organ.

Fuchu He, PhD, Member of the Chinese Academy of Sciences, Member of the Academy of Sciences for the Developing World. He received his B.S. degree in Genetics from Fudan University, Shanghai in 1982. Then he earned his M.S degree in Biochemistry and Ph.D. in Cell Biology from Beijing Institute of Radiation Medicine. He is currently the Director and Professor of State Key Laboratory of Proteomics, Beijing Proteome Research Center and Beijing Institute of Radiation Medicine; the Director of China National Center of Biomedical Analysis; council member of Human Proteome Organization (HUPO); co-chair (inaugural chair) of HUPO Human Liver Proteome Project (HLPP), the vice-president of AOHUPO, the president of CNHUPO. His major fields of research are proteomics, genomics, bioinformatics and systems biology, especially for liver physiology and pathology. Prof. He holds the important positions for the leading proteomic journals, including senior editors of Proteomics and Proteomics—Clinical Application and editorial board member of Molecular & Cellular Proteomics.
As the chief scientist or principal investigator, Prof. He has been awarded several important research grants, including those from the National Basic Research Program ("973"), the National High Technology Research and Development Program (“863”), and the National Natural Science Foundation of China (NSFC). He has published more than 130 peer-reviewed papers in SCI international journals as corresponding author, including Nature Cell Biology, Nature Genetics, Nature Biotechnology, Nature Protocol, PNAS, EMBO J, Gastroenterology, Hepatology, Genome Res, Mol Cell Proteomics, Cancer Res, J Proteome Res, JBC, Oncogene, Proteomics, Bioinformatics, etc.