3D Gene Expression

Expression Atlases
Search PointClouds
Browse PointClouds
Data Releases

Overview:
   Summary
   Project Goals

Methods:
   Staining & Mounting
   Probe Constructs
   Staging
   Imaging

Software:
   PointCloudXplore
   PointCloudToolBox
   PointCloudAlign
   NODE

File Formats:
   PointCloud (.pce)
   VirtualEmbryo (.vpc)
   Correspondence files

Image Analysis: PointCloudToolbox

Each of the 0.3-0.5 Gb, 3D confocal image stacks is processed and analyzed to obtain a 1 Mb text file that we call a PointCloud. This file contains geometric data for each nucleus (position, volume, etc.), and the level of expression of the two genes. The gene expression levels are measured independently in the nucleus, and in both the basal and the apical regions of the cytoplasm associated with each nucleus.

The set of image analysis algorithms we employ use standard and not-so-standard techniques and address a series of problems, including signal attenuation, channel cross-talk, noise and the lower resolution along the optical axis (which is about 3 times lower in this axis than in the optical plane). This last issue requires the algorithms to estimate locations of some nuclei because boundaries between nuclei largely disappear in some areas of the image.

PointCloudToolbox is a MATLAB toolbox that contains the routines to produce PointCloud version 5 files, as well as routines to aid in analyzing morphological and gene expression features. It requires MATLAB 7 with the DIPimage toolbox, and is not intended to be used by people that don't want to get their hands dirty. Please read the online user guide for installation and usage instructions.

Additionally we provide an older version of the routines, that generated the PointCloud files used in Luengo et al, 2006, and Keranen et al, 2006 (PointCloud version 4).

The part of PointCloudToolbox that produces PointCloud files is also available as a standalone executable (pcmaker). With it you can generate PointCloud files from a list of embryo images, as well as all the data files required to upload those PointCloud files to our internal database, BID. This program does not require MATLAB nor the DIPimage toolbox. Please read the online user guide for installation and usage instructions.

The standalone executable contains DIPlib and other portions of the DIPimage toolbox, ©1995-2007 the Quantitative Imaging Group, Delft University of Technology.

The standalone executable contains code from MATLAB, ©1984-2007 The MathWorks, Inc.

PointCloudToolbox and the standalone executable contain the EigenCrust program, ©2004-2007, Ravi Kolluri, University of California at Berkeley.