LifeSKIM is a highly customizable and scalable application that offers a novel way to explore the existing public and in-house biomedical knowledge. Biological databases provide vast amounts of structured information that is heavily interlinked with unstructured or semi-structured textual documents. This unstructured information locks some of the most valuable knowledge because it is presented in a non-formal way. In order to have a transparent knowledge access, we need to generate a critical mass of semantic annotations between the structured and the semi-structured information (documents).

The LifeSKIM application has the following main features:

• Semantic indexing and retrieval of documents using knowledge bases
• Flexible querying and navigation of knowledge, generated from both structured and unstructured text
• Recognition and population of new entities from text
• Efficient reasoning against the extracted and structured information
• Co-occurrence and ranking of entities in documents
• Classical information retrieval tasks
• Persistence and annotations of arbitrary data types

The LifeSKIM knowledge base contains a sophisticated terminology vocabulary composed from more than 1'000'000 biomedical entities. These entities were extracted from MeSH, SNOMEDCT CT, ICD-10, Gene Ontology, Taxonomy, Entrez-Gene, and DrugBank data sources. To align the different terminologies in a semantically consistent dataset and introduce new formal relations we have used UMLS. For instance "Barrett's esophagus" is a "disease of gastrointestinal tract mucous membrane" and a "soft tissue lesion". The LifeSKIM knowledge base can be extended with additional information sources like PIKB or in-house databases.

The LifeSKIM information extraction pipeline uses a state of the art technology that identifies and maps biomedical entities to the LifeSKIM knowledge base. It is able to recognize nearly a half-million entities such as: genes/proteins, diseases, cell lines, cell types, organisms, drugs, DNA, RNA, subcelular localizations, cellular components, biological processes, molecular functions, etc. The list of recognized entities can be extended with standard information extraction pipelines developed by the General Architecture for Text Engineering (GATE) platform. The LifeSKIM application features natural language processing algorithms you can use out of the box or modify for specific in-house solutions.

To access the public LifeSKIM application server, please send a request to lifeskim[-at-]ontotext.com

To get a quick introduction, please check our demonstration movie