Worm Breeder's Gazette 14(5): 30 (February 1, 1997)

These abstracts should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.

Visualizing the Embryogenesis of C. elegans

Hiroaki Kitano

Sony Computer Science Laboratory 3-14-13 Higashi-Gotanda, Shinagawa Tokyo 141 Japan

When Sydney Brenner proposed a new project to investigate  C. elegans
to the Medical Research Council, he chose it because it is the
simplest possible differentiated organism. The decision was right, and
now C. elegans is the most well investigated multi-cellular organism.
Since the proposal to use this organism as a model organism, a series
of research projects on various aspects of this organism were
inititated. As a result, the complete cell lineage, neural circuits,
and various genes and their functions were identified, and the
complete DNA sequencing  and the gene expression for each cell at
different times in the embryogenesis will be identified in a few years. 
Despite the fact it is the simplest possible differentiated organism,
it is still too complex to understand the dynamics and interactions
taking place.  Given the abundance of data, we consider that introducing
a synthetic approach will further enhance our understanding of the
underlying principles of development and behavior of C. elegans. We have
started a project which we have named the Perfect C. elegans project,
which aims at implementing detailed model of C. elegans on a computer
system.  As a first step, we have developed a computer graphics
visualization system of embryogenesis on C. elegans. The system is based
on existing data on the development of C. elegans, and missing
information is interpolated using a simulation technique. The current
system generates computer graphics images of the embryogenesis of C.
elegans up until 500 minutes after the first cell division. 

The three dimensional (3D) visualization system is an appropriate
starting point because it provides a 3D model of C. elegans, so
that the cell-cell interaction dynamics, at both the physical and
chemical level, can be implemented on top of this model. This would
greatly help research on development. The current system is based on
the cell lineage  and cell location data published in Sulton's papper
published in 1983. We are also working on the newer data from the
Sanger Centre.

It is a non-trivial task to create a reasonably accurate computer
graphics image based on the available data because information
necessary to create three dimensional models is missing. Following
information were available: (1) the complete cell lineage chart, (2)
hand-drawing pictures in 2-1/2 dimension (all 28 cells at 100 minutes,
55 out of 180 cells at 200 minutes, 137 out of more than 350 cells at
260 minutes, 156 out of more than 350 cells at 270 minutes, most cells
at 430 minutes), qualitative descriptions of the shape of embryo,
qualitative description of disparity in the size of divided cells, and
general information on migration. 

In order to create an accurate computer graphics images, we need three
dimensional data on the position and shape of the cell in a series of
time steps.  Obviously, such data is not available. The challenge is
to how estimate reasonably accurate cell position data from available

Our strategy to overcome this problem is to merge simulation with
data. First, in order to assure the accuracy of the computer graphics
image, cells must be in the position given in  the observed data.
Second, various simulation techniques are used to fill in the missing
information, such as the location of cells not provided in the data.
Essentially, this part of the system computes forces between cells,
such as the force which pushes back colliding cells (equations not
shown). However, if only dynamic simulation is used to decide the
position of cells, some cells will not be in the position described in
the observed data, because of cell movement. In order to compensate
for this discrepancy, the force that a  cell is supposed to generate
for its movement was estimated using inverse kinematic techniques, and
added to the cell's force vector. The motions of the cell are computed
as in the case of the motion of objects in a viscous fluid. 

Along with other simulation techniques, the system creates reasonably
realistic 3D computer graphics animation image from the first cell
division to about 500 min after the first cell division.
Cells are colored by their cell fate, or by their founder cells.
Due to the animation capability, movement of the cell can be clearly
identified, and helps intuitive understanding of the behaviors of cells
during embryogenesis.

We are now working on more accurate simulation using newer data sets,
as well as implementation of genetic information into the simulation.