Knowledge Bank

SigNET Knowledge Bank – Annotated and Predictive Information About Proteins, Their Regulation and Interactions.

The SigNET Knowledge Bank is a suite of open-access, online websites and databases developed by Kinexus Bioinformatics Corporation to foster research into cell SIGnalling NETworks to advance biomedical research in academia and industry. This resource has been produced to aid researchers in their studies of human protein mutation (OncoNET), expression (TranscriptoNET), interaction (KinATLAS, DrugKiNET) and protein phosphorylation (PhosphoNET, KinaseNET) to aid in the design of new tools and experiments, and the interpretation of experimental results. Additional knowledgebases are currently under construction to provide detailed information on receptors, protein kinases, protein phosphatases, adapter, stress and transcription proteins. With the launch of KinATLAS later this year, it will be even easier to track known and predicted protein-protein and kinase-drug interactions. We hope that the SigNET Knowledge Bank will contribute significantly to the development of predictive, systems-based proteomics research.

The TranscriptoNET KnowledgeBase is an open-access, online resource that features comprehensive information on the mRNA expression levels of about 21,000 genes in about 600 types of human organs, tissues and cells as monitored with gene microarrays. The original data used in TranscriptoNET was retrieved from the National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO), which serves as a repository of experimental gene microarray results submitted by diverse academic and industrial laboratories around the world. With the aid of its academic collaborators, Kinexus has normalized the data from over 900 different studies with over 6000 biological specimens to permit investigations of gene expression and potential interactions that can only be undertaken with such a large dataset of over 125,000,000 gene expression measurements. This normalization process was based on the identification of 60 genes that were commonly and highly expressed in all of the biological samples.

In the selection of human specimens for inclusion in TranscriptoNET, special attention was paid to human tumours and cancer cell lines to identify the differential regulation of genes in cancer. This may uncover new potential oncogenes and tumour suppressor genes that may encode cancer protein biomarkers and drug targets. We invite the biomedical research community to use TranscriptoNET as a powerful tool for discovery of genes that are uniquely or commonly expressed throughout the human body, and to discover possible functional interactions amongst the 21,000 proteins encoded by the human genome based on their co-expression patterns. The differential expression of genes determines the structures and biochemical activities in cells that define their physiological functions. TranscriptoNET could also be used to aid researchers in uncovering how the body’s diverse organs, tissues and cells may be developmentally related.

Soon to be released, the OncoNET KnowledgeBase provides detailed information on differential mRNA expression and mutation for about 3000 human genes in diverse types of cancers. The gene expression data was derived from the NCBI GEO database, and most of the gene mutation data was annotated from the Sanger Institute’s COSMIC database. OncoNET is a powerful tool to guide cancer researchers in the identification of potential biomarkers and therapeutic targets for cancer diagnosis and treatment. The cancer-related proteins profiled in OncoNET are broadly divided into three categories: those that facilitate cancer, which are known as oncoproteins (OP) or tumour requiring proteins (TRP); and those that inhibit cancer, which are classified as tumour suppressor proteins (TSP). A unique combination of gain-of-function of OP and loss-of-function of TSP underlie each human cancer. By contrast, TRP are rarely mutated. The alterations in the activities of OP and TSP may arise from mutations, altered gene expression, and the actions of environmental agents that directly or indirectly impinge on these proteins.
The PhosphoNET KnowledgeBase presently holds data on more than 180,000 empirically-determined phosphorylation sites in about 16,000 human proteins as well as another 780,000 putative sites that have been identified with Kinexus’ proprietary phosphosite prediction algorithms. The identities of experimentally-confirmed phosphosites and information about specific protein kinases and phosphatases that target many of these sites were obtained from collaborators, the scientific literature, and other reputable websites such as PhosphoSIte Plus, PHOSIDA, UniProt and PhosphoElm.

The vast majority of human phosphosites on proteins are unlikely to exert direct functional effects, although they may facilitate protein degradation. Functional data is available for only about 2500 known phosphosites, most of which are documented in PhosphoNET. To help discover other promising functionally-relevant phosphosites, PhosphoNET features an Evolutionary Analysis Module that explores the conservation of 960,000 known and suspected human phosphosites in 22 other diverse species. The most conserved phosphosites in across evolution are likely to play important roles in regulating cellular processes.

At least 538 protein kinases catalyze the phosphorylation of 21,000 diverse human proteins. Kinexus and its collaborators have developed kinase substrate prediction algorithms that have scored the ability of 500 candidate kinases to phosphorylate each of the 960,000 confirmed and putative human phosphosites. The Kinase Predictor Module in PhosphoNET lists the top 50 ranked protein kinases for each phosphosite. Additional predictions of protein phosphatase and adapter protein interactions with human phosphosites will be added to PhosphoNET in the future.

KinaseNET with detailed information about the 536 known human protein kinases is another SigNET KnowledgeBase that shall be released shortly. KinaseNET includes comprehensive data on protein kinase structures, regulation, specificities, substrates, distribution, evolutionary conservation and inhibitors. However, the most complete data on the inhibition of protein kinases by small molecules is provided in our DrugKiNET KnowledgeBase. DrugKiNET features comprehensive information on over 800 compounds that have been experimentally determined to inhibit human protein kinases. This includes the retrieval of the lowest reported compound IC50, Ki and Kd values from several sources, including the National Center for Biotechnology Information (NCBI) PubChem Compound database, the Kinase SARfari database from the European Molecular Biology Laboratory (EMBL) European Bioinformatics Institute, The International Centre for Kinase Profiling at the University of Dundee, Ambit Biosciences and hundreds of original research publications. Using over 105,000 experimentally tested, non-redundant kinase-compound pairs for training, we have developed two kinase inhibitor prediction algorithms to further predict another 200,000 kinase compound interactions. DrugKiNET can also provide insights into the kinase amino acids that are critical for drug interactions.

Later this year we will be launching KinATLAS, which will generate maps of protein-protein and kinase-drug interactions, with extensive hyperlinking to the other SigNET Knowledgebases as well as external websites such as UniProt. Unlike other pathway mapping websites, KinATLAS will provide indications of the predicted strength of interactions and the levels of proteins in a cell/tissue-specific manner.

For updates on the release of KinATLAS, OncoNET, KinaseNET and other upcoming SigNET knowledgebases, please register with our Mailing List and Twitter.

Click here to open TranscriptoNET

Click here to open PhosphoNET

Click here to open DrugKiNET