current projects‎ > ‎

MelanoBase (SNF)

The main aim of the MelanoBase project is a  large-scale automatic extraction of actionable information from the biomedical literature and its integration with existing structured knowledge (life science databases).  The innovative outcome of this strategy is to provide users  (basic and clinical researchers) with formats that can be more easily queried, and automatically processed, with the purpose of increasing the efficiency of biology research. The specific use-case scenario of melanoma disease has been selected for the histopathological complexity of these lesions, and to provide solutions for the unmet need of separating true drivers of this disease from a myriad of (epi)genetic inconsequential byproducts accumulated during melanoma genesis. The project will pursue a literature-wide and disease-centric approach which sets it apart from comparable projects worldwide. Moreover, close collaborations with experts in the field will streamline validation efforts in clinically-relevant specimens.
The importance of knowledge integration across the life sciences, including literature processing, is emphasized by the number of research programs which focus on this goal. For example, the DARPA-funded "Big Mechanism" initiative aims at building computer models of cancer mechanisms using information obtained from automated reading of research papers. The German initiative i:DSem (Integrative Datensemantik in der  Systemmedizin) has the goal to promote the development of medicine through the integration of structured and unstructured data sources, including literature processing.
The MelanoBase project aims at integrating all available knowledge about melanoma, with particular emphasis on hard-to diagnose lesions and on mechanisms of resistance to clinically approved treatments and compounds in experimental testing.   The resulting knowledge resource will be tested in  the context of some European leading cancer research centers, and a large pharmaceutical company. Melanoma provides a relevant testbed for experimentation with large-scale literature analysis, not only for the obvious societal relevance of the disease, but also because it represents a prototype of inherently complex and variable tumors. All known oncogenes and tumor suppressor programmes known to date are directly or indirectly deregulated in melanoma.
Thus, with over 80,000 mutations described, and plethora of post-transcriptional changes, the amount of interconnections between genes, and altered  phenotypes or signalling cascades is so challenging, that a thorough analysis  would require a colossal amount of experimental research.  High-throughput literature mining can provide useful clues to help prioritize candidate targets, on the basis of evidence from previous experiments.
The primary goal of MelanoBase  is to enable integration of the unstructured knowledge available in the literature with the structured knowledge deposited in life sciences databases. Additional sources such as, for example,  clinical trial reports, systematic reviews (Cochrane), and prescription drug information might also be mined in a second stage of the project.  Our ultimate goal is accelerate gene discovery and drug target validation in the area of melanoma.
The results of the MelanoBase project will be integrated  within the Melanoma Molecular Map repository and experimentally relevant information will be validated by well-known experts in the field.

Duration: 3 years (2016-2018)
Funding: CHF 420'469, Swiss National Science Foundation (grant CR30I1_162758)
Principal investigator: Dr. Fabio Rinaldi

Main collaborations:


  • In October 2016, we started a strategical collaboration with the HTL-NLP group at the Fondazione Bruno Kessler (FBK), an AI research institute based in Trento, Italy, with a long and prestigious tradition of research in natural language processing (NLP) and machine learning (ML). We expect a significant contribution to the development of advanced approaches for knowledge extraction from the scientific literature.