Delete search term

Header

Quick navigation

Main navigation

Computational literature-based natural product drug discovery

Description

Natural products such as medicinal plants and extract mixtures are an important source of pharmaceuticals. In 2015, Tu Youyou was awarded the Nobel prize in medicine for her discovery of artemisinin isolated from a natural product as a treatment against malaria, saving millions of lives.

Medically relevant substances and their properties are often found through systematic, time-consuming literature analysis, nowadays referred to as Literature Based Discovery (LBD). In his pioneering work in the 1980s, Don Swanson found hidden links between unrelated pieces of knowledge in the scientific literature. He found, for example, a publication showing that fish oil (A) can reduce vascular reactivity (B), and a different publication showing that a reduction in vascular reactivity (B) can treat Raynaud’s syndrome (C). Such knowledge can be linked. From A➝B and B➝C one can deduce A➝C, i.e. consumption of fish oil could benefit patients with Raynaud’s syndrome. Swanson hypothesized this novel link, which was later confirmed through clinical trials.

Although Literature Based Discovery (LBD) has become widespread, little has been done to automate it in the field of Natural Product Drug Discovery. The common practice is manual text mining, which is extremely laborious and time-consuming.

The goal of the project is to devise an automated LBD system for natural product drugs. The system can be used not only for new discoveries, but also to query and explore present knowledge. Further, it reduces discovery costs by saving time in evaluating literature for drug development and is also relevant to drug safety aspects.

The project is an interdisciplinary collaboration between natural products research (Andreas Lardos, Evelyn Wolfram, ICBT) and computational science (Maria Anisimova, Manuel Gil, IAS). In addition, a PhD student and Master students from the Applied Computational Life Sciences specialisation at ZHAW will contribute to the project.

Graphic

The discovery system operates on a large body of scientific literature relevant for natural product drug discovery. This natural language text is converted to a formal representation (steps 1 and 2). Ontologies provide a common vocabulary, machine interpretable definitions of concepts, and relations among them. They are aligned (step 3) with the formal representation of the literature. Given a user query, the resulting semantic graph (or network) is explored by discovery algorithms.