FlyTED project
From ImageWeb
FlyTED: The Drosophila Testis Gene Expression Database
First draft - needs review and refinement - GK
This web page describes part of a larger project led by Dr Helen White-Cooper investigating gene expression in Drosophila fruit fly testes.
FlyTED, the Drosophila Testis Gene Expression Database, is a public database containing mRNA in situ images and ancillary data revealing the extent of expression of individual genes involved in spermatogenesis in the Drosophila melanogaster testis, both in wild type flies and in a number of meiotic arrest mutant strains.
This dataset was generated by Elizabeth Benson, Elin Gudmannsdottir and Helen White-Cooper in the Department of Zoology at the University of Oxford, with funding from the UK's BBSRC. The database system has been developed by Jun Zhao of the Image Bioinformatics Research Group in the Department of Zoology at the University of Oxford, with funding from the UK's BBSRC.
Duration: December 2004 to December 2007.
The database is hosted at http://www.fly-ted.org.
Contents |
Project goals
The aim of this project is to make a substantial dataset of gene expression images, along with associated metadata needed for their interpretation (Drosophila strain, gene identifier, anatomical region of expression) publicly available on the web. The intent is that this data may be used, with other web data sources, to provide a basis for full review of published results, and to further the work of future research projects in the same area by contributing to the publicly accessible fund of information about gene expression patterns.
Project focus
The database aspect of the FlyTED project is adapting an existing OAI repository system, EPrints from Southampton University, to create a web-accessible database of gene expression images and their associated metadata.
The user interface aspect of this project is to provide a direct web access to the images and data, based on the existing EPrints user interface.
The longer term goals, which will be pursued by future research projects, include:
- using Semantic Web standards and technologies to provide machine-processable access to the images and data to facilitate combination of these with other web-accessible data sources;
- using Semantic Web standards and technologies to provide alternative and more flexible user interfaces over these images and associated metadata. Tools that we have inspected include jSpace and Exhibit. And tools that are in our agenda include mSpace and other Semantic Web applications.
Functional goals
- To create a web site that can be used to view a broad range of images and annotations created by Dr Helen White-Cooper's team at Oxford University.
- To provide a basis for programmatic access to the data so that it can be combined with other data sources, or further processed by external software applications.
Project activity
Discontinued directions
Under the banner "Bioimage", Our research group has been examining the requirements and technology for web publication of research images and data for several years. Initially, we tried a conventional (relational) database approach, which worked well enough but was insufficiently flexible, and was unable to take account of known relationships between possible query terms (e.g. that all rats are also mammals).
A second generation image database was based on semantic web tools and standards, which hold the promise of flexibility, evolvability of data, merging information from multiple sources and exploiting knowledge about the things described. We did have considerable success with this, building a database using Tomcat, Java servlets, Java Server Pages, STRUTS, Sitemesh, Jena and Postgress. But in the end, the software was very susceptible to break when new versions of the supporting tools were released, and the maintenance load was more than a small software team could stay on top of.
This leads us to the current approach, where we aim to use existing, supported tools to perform the "heavy lifting", and focus our efforts on linking them together in ways that allow the underlying components to evolve separately, each at their own pace.
Current work
Current work includes the following activities:
- Adapting the EPrints database to hold annotations for displayin the standard EPrints web interface
- Writing custom code to load images and annotations, using the researchers' original Excel spreadsheets
- Adapting the EPrints web interface templates to display the image thumbnails and additional annotations
- Adapting the EPrints search interface templates to display the additional annotations
Custom code has also been written to extract images and metadata using OAI-PMH and HTTP protocols, but that is really part of another project (FlyWeb, FlyData).
Project meetings
- 29th June, 2007 Kick off with with Helen
- 09th July, 2007 First demonstration to Helen's group
- 16th July, 2007 User requirement meeting with Liz and Elin
- 17th August, 2007 Feedback meeting with Helen
- 9th October, 2007 Feedback from Liz and Helen
Technical notes
Image browsing
The visual nature of images, and the way that people deal with them, that summary display benefits from a rather different approach to that used by the default EPrints display. Instead of presenting an entry-per-row for images and summary information, the main browse interface is a grid of thumbnails with minimal additional information.
Storing annotations in EPrints
The annotations are loaded twice:
- Once into the EPrints database for use in the EPrints web interface. This might be a subset of the available annotation data.
- Once into a date stream that is part of an EPrints record containing one or more images; the content of this data stream is opaque to EPrints itself, but canj be harvested by specialized extraction tool using a combination of OAI-PMH and simnple HTTP requests.
Continuation and sustainability
The current work has involved direct modifications to some of the EPrints code and templates. For sustainability, we would prefer that the EPrints software can be used with minimal modification, so that our developments can be used with new versions of EPrints as they are released. To this end, we would like to be able to use standard EPrints plugins for importing and exporting images and metadata; e.g.
- METS for importing images and annotations
- METS and/or OAI-PMH for exporting images and annotations
It is less clear how we might isolate the web user interface changes from ongoing EPrints development.
In order to display annotations through the EPrints web interface, it has been necessary to extend the EPrints database schema. This makes the deployment very specific to our application. We have wondered if any of the following might be possible:
- a common database schema for storing named attributes with a record
- a plugin mechanism that can retrieve domain-specific metadata from data streams associated with a record (e.f. RDF or Atom), and use these as part when generating EPrints web pages.
Related work
- http://fly-fish.ccbr.utoronto.ca/ - Fly-FISH, a database of Drosophila embryo mRNA localization patterns. This database documents the expression patterns of Drosophila mRNAs at the subcellular level during early embryogenesis.
Other notes and resources
Links
-- GrahamKlyne 12:07, 11 September 2007 (BST)
