BioTurk: Crowdsourcing the Construction and Augmentation of Biological Pathways

Abstract:

A biological pathway can be modeled as a network of sparsely connected proteins, where network paths correspond to a sequence of protein-protein interactions that lead to higher-level changes within the cells of the human body. a deeper understanding of how proteins interact to carry out biological goals can lead to insights in human disease research an the proposal of new drug targets. However, the number of pathways we currently understand is limited, and these pathways are incomplete. Although experimental evidence for many protein interactions and protein=-pathway affiliations exist in research literature, relatively few have been manually curated.

We propose using crowdsourcing to aggregate published biological research in order to construct new pathways and augment existing ones. We hope to use these pathways to enable accurate prediction of yet more pathway components and drug targets based on functional genomics data. We developed a web application (http://bioturk.princeton.edu) where untrained users can answer simple biological questions by using their ability to deconstruct sentences. By deciphering sentences from PubMed, these users will help us support biomedical research. AT the same time that the users learn more about biology and the biological pathways relating to disease, our approach can also leverage participation by expert biologists. These biologists are given contested questions, have their answers up-weighted, and utilize an expert user interface that allows them to provide corrections to the pathways.