Abstract

46th Annual Drosophila Research Conference, San Diego, California, March 30-April 3, 2005

Visualization Tools for Three-dimensional Gene Expression Data in Drosophila.

G.H. Weber1,2, C.L. Luengo Hendriks2, S.V.E. Keränen2, S.E. Dillard1, B. Hamann1, Berkeley Drosophila Transcription Network Project.

1) Institute for Data Analysis and Visualization, University of California, Davis, CA.

2) Genomics and Life Sciences Divisions, Lawrence Berkeley National Laboratory, Berkeley, CA.

It is intrinsically difficult to visualize complex gene expression patterns and morphology in 3D, especially at cellular resolution. As part of the Berkeley Drosophila Transcription Network Project (BDTNP), we are developing methods to overcome this challenge. Specific tools under development permit 1) visualization of raw confocal microscopy data, 2) interactive control and verification of the image processing algorithms and 3) visual analysis of complex gene expression patterns. By adapting novel volume-rendering methods to multi-channel confocal microscopy data, every single channel of an image stack is rendered independently by mapping brightness information to color and transparency information. Resulting colors are blended using user-specified weights, supporting seamless blending between channels. Three-dimensional slicing and rotation tools make it possible to visualize an arbitrary cross-section from an image stack, and to explore the interior of an embryo. Volume visualization has been combined with user interaction tools to support quantitative determination of the accuracy of the BDTNP's nuclear segmentation methods. Nuclei are rendered to include information obtained from a nuclear segmentation mask (see abstract by Luengo Hendriks et al.). It is possible to select individual nuclei interactively and identify falsely segmented objects. This work has already yielded significant improvements in segmentation accuracy. The BDTNP’s final 3D gene expression data is in the form of matrices or “point clouds” that describe nuclear positions and their associated gene expression levels (see Luengo Hendriks et al.). We are developing a point cloud visualization tool that will be offered as an effective means for the fly community to access and analyze 3D expression data from the BDTNP.

Last modified April 9, 2005.