Building Composite Maps of Gene Expression Patterns: Registering Morphology and Gene Expression Between 3D Representations of Drosophila Embryos.
1) Computer Science Division, University of California, Berkeley, CA.
2) Life Sciences and Genomics Divisions, Lawrence Berkeley National Laboratory, Berkeley, CA.
Spatial patterns of gene expression in animals are typically analyzed by visual inspection of photographic images produced by in situ hybridization experiments. A profound limitation of this approach is that it captures little 3D structure and provides only rough quantitative information about expression patterns. The Berkeley Drosophila Transcription Network Project is developing a suite of methods to convert image stacks generated by confocal microscopy into computer searchable representations of gene expression at cellular resolution. One key difficulty is that fluorescence microscopy can only capture expression levels for a few (2-4) gene products in a given animal. Here we report a method for registering 3D expression data from different Drosophila embryos stained for overlapping subsets of target genes in order to build a composite map, ultimately containing collocation information for 1000s of genes. Each embryo image stack is segmented into cellular volumes with the help of a nuclear stain and the fluorescence level quantified in each volume (see posters by Keränen et al. and Luengo Hendriks et al.). This yields a table of 3D nuclear/cellular locations along with their associated expression levels. To combine such data from two or more embryos, it is necessary to identify corresponding cells in each embryo and then transfer the associated expression levels. Using tools developed in computer vision, we formulate this as a 3D deformable template matching problem and show how to overcome significant distortions in the shapes of embryos as well as biological variation in morphology and differing numbers of cells. We demonstrate composite maps of late-stage-5 blastoderm showing collocation of several early patterning genes.
Last modified April 9, 2005.