NeuroMorpho.Org: An exemplary neuroinformatics resource for cellular neuroscience.

摘要

Neuroinformatics is defined by the adoption of computational approaches and analytical tools to facilitate neuroscience data management and processing. Following its start two decades ago in response to the massive outburst of heterogeneous neuroscience information, neuroinformatics has substantially contributed to recent research progress. Web-accessible digital resources, enabling electronic access to on-line experimental and model results, constitute core elements of neuroinformatics by catalyzing the transformation of "data" to "knowledge". Successfully creating and maintaining such resources are technically challenging, time consuming, and costly. Identifying the limiting factors, sensible solutions, and best practices in this process is very important.;This dissertation describes the development and maturation of NeuroMorpho.Org as an exemplary neuroinformatics success story. NeuroMorpho.Org is the largest repository of 3D digital reconstructions of neuronal morphologies. Since morphology is a key determinant of neuronal function, quantitative neuromorphological data are crucial to understanding structure-activity relationships in the brain. 3D digital reconstructions are computationally parsimonious representations of neuronal morphologies and may take several weeks each to acquire experimentally. NeuroMorpho.Org contains >5000 such reconstructions from 39 labs worldwide, 11 animal species, 15 brain regions, and 40 cell classes. The site offers dynamic browsing and searching based on metadata, morphometrics content, and keywords. To date, more than 20,000 visitors from 68 countries have downloaded over half-million files. NeuroMorpho.Org is also linked to an extensively annotated custom-designed database of peer-reviewed literature reporting digital reconstructions.;This research showed that leveraging the power of common computational technologies in a flexible and interoperable fashion, choosing a valuable and well-defined data type, providing dense coverage, and actively accelerating data sharing, all extensively contribute to success. After covering ∼65% of all relevant publications by extensive literature mining, follow-up communication with authors of the 698 positively identified articles indicated that only 24% of the reported reconstructions will eventually become available for sharing. The accessible NeuroMorpho.Org content is representative of these data. More than 70 papers published in relation to NeuroMorpho.Org, including modeling studies, comparative studies, and studies of the cellular elements required in reverse-engineering the brain, vetted this resource as a reliable test-bed for developing new tools, encouraging scientific collaboration, and fostering neuroscience discovery.