A major obstacle to translational research is the formulation of hypotheses that span multiple levels of biomedical research. This is particularly true for complex neuropsychiatric disorders, which involve aspects of many scientific disciplines at many physical scales. Combined efforts involving multiple disciplines are needed in order to make any substantial progress. We propose development of a software platform called the Hypothesis Web that permits collaborative formulation of complex scientific hypotheses. It represents a novel approach to the way scientific hypotheses are conceived and tested, in which: (1) hypotheses are the focus of collaboration; (2) literature, data, and annotations are organized in a usable interface; (3) translational research groups work together to develop a web site for the hypothesis that contains accumulated evidence; (4) hypotheses that span multiple levels are represented as hierarchical graphs; (5) hypotheses can be inspected analytically with tools like meta-analysis; (6) resulting hypothesis web sites can be shared, modularized, and published. We propose to develop this platform as an extension of existing open-source systems for web-based collaboration, data management, modeling, and access of online scientific literature. Specifically: a wiki-like platform augmented for collaborative refinement of hypotheses. The Hypothesis Web represents hypotheses visually as labeled graph models that identify the relevant concepts and associations. This graphical approach permits visual specification of hypothesized models of different kinds, including pathway diagrams, measurement models, and structural equation models. The graphs can then be represented in 2D cartographic displays that we refer to as Hypothesis Maps. Our design for the Hypothesis Web builds on many results for phenotype analysis developed under our P20 planning grant (P20RR020750). The system will be used to record hypotheses developed by Consortium Projects. It will be developed as an extension of existing open-source systems, and be made freely available as it evolves over the duration of this project. We also propose development of a platform for producing a Phenomic Atlas, using the Hypothesis Web as a foundation, but extended with knowledge-based capabilities for exploration. Conceptually a related set of Hypothesis Webs can be bundled together to form an Atlas. Thus the Atlas is a medium for publication that permits dissemination of a collection of related Maps (hypotheses). In collaboration with Consortium Research Projects we will construct Phenomic Atlases.
Showing the most recent 10 out of 14 publications
