Note: Because much of the work discussed centered on current research and unpublished results only
very brief summaries of the material are presented. In some case references to or results taken from published
results are given, but otherwise no original research actually presented at the meeting is shown here. It is to
be emphasized that this was a workshop and not a research symposium.
Elliot Meyerowitz provided some
introductory comments and gave a brief definition of computational morphodynamics, and mentioned some of the
groups doing research in this are such as the computable plant
project.
Marcus Heisler (Meyerowitz Lab) talked about
linking biochemical signalling with mechanical models to develop a more comprehensive description of growth
patterning in the shoot apical meristem. He obtains dynamic
images of cell division and movement via a laser scanning confocal microscope and utilizing various Green Fluorescent Proteins (GFP's). Of particular
interest is the development of microtubules that form the cytoskeleton, such as those described by the Shaw Lab at Indiana; the importance of Auxin
for cell signalling as described by Reinhardt's Group in Switzerland; and studies
of meristem architecture being performed by Jan
Traas' group in Lyon.
David Sprinzak (Michael Elowitz lab)
discussed the development of fine grained differentiation patterns during development in terms of a framework in
which networks are reconstructed in
mammalian cells [Sprinzak & Elowitz,
Nature 438:443-8 (2005)].Synthetic circuits can be used as simple in vivo models to explore the relation
between the structure and function of a genetic circuit. An example of notch-delta signalling was given,
similar to the model of
Collier et al. [JTB 183:429-46 (1996)] which reconstructs various checkerboard-like patterns.
Melissa Pope, Ehsan Jabbarzadeh, and Keiichiro Kushiro (Anand Asthagiri group) described how various aspects of
cell adhesion affect epithelial patterning during development. The physical interactions of a cell with its
environment includes both cell-cell interactions and interactions with an an extracellular solid-state matrix
of proteins. In addition to providing a physical platform, cell-matrix adhesion also stimulates various
biochemical signals and networks. Their lab is engineering physical platforms that elicit similar control over
cell behavior and using them to study mammalian cell behaviors and pattern development.
Eric Mjolsness discussed the
need for morphodynamic mathematical frameworks. A large number of frameworks have been proposed and
successfully used to described various aspects of morphodynamics, including: generalized reaction networks;
fixed cells; Turing models; PDE's that describe cell growth and polarity; cellular compartmental models
including weak spring systems, finite element descriptions, and lively cell complexes; spatially stochastic
approaches; and unified approaches such as variable structure grammars. [for a review see Mjolsness,
J Plant Growth Regulation 25:270-277 (2006)]. He also proposed a definition of computational morphology:
the attempt to answer the question, "How do biochemical and informational processes determine major changes
in the morphology of living organisms?"
Alex Cunha of the Caltech Center for Advanced Computing
Research summarized many of the image processing techniques that are being used on campus to do image
analysis; filtering; segmentation; and de-noising. He referred to the paper by [Buades, Coll, & Morel,
Multiscale Modeling and Simulation, 4:490-530 (2005)] as providing some standard algorithms.
Boris Shraiman (UCSB/KITP)
discussed the relationships between mechanics, growth, size and pattern control. He talked about how morphogen
gradients affect the development of the fly wing [PNAS 104:3835-3840 (2007)] and the importance of
mechanical feedback as a growth regulator [PNAS
102:3318-3323(2005)]. During the discussion he also posed a question for us in our research: are we looking
for universality or computational efficiency (in our software/analysis/models)?
Greg Reeves (laboratory of Angela Stathopoulos discussed
projects in which he is studing the spatial gradients of the dorsal protein in drosophila. The development of
the dorsal/ventral protein in drosophila occurs as a result of a gradient that develops in the nuclear
transcription factor called Dorsal. Dorsal is a maternally deposited rel-containing transcription factor that
is present in a nuclear gradient within the early Drosophila embryo.
Ingmar
Riedel-Kruse discussed his studies of synchrony dynamics of the segmentation clock in drosophila and
how it affects patterning in the embryo. [ref: Science 317:1911-1915].
The following gives a summary of points that were brought up during the final discussion.
- To understand shape formation one needs to understand (among other things): (1) cell/cell interactions
and (2) cell/substrate interactions.
- This is a diverse and widely multidisciplinary field. Is there a way to turn some of this information
into some sort of course. We need focused short courses that will help people understand enough of the
analytic/computational techniques to be able to work with a modeler - much less be able to do our own
modeling.
- We need to have more interactions like this and sharing of information. It would be useful to have:
journal clubs; multi-lab group meetings focuses on morphodynamics; tutorials on image analysis. It was
decided that we would start having monthly super-group meetings with this aim.
- Exchange of knowledge is essential. We need to know where to get information: even simple web pages with
lists of course, books, articles, software, etc, that tell us where to go to find things out will help a lot.
The BNMC offered to post the information that people suggest.
- Is common tool development possible? There is not a single common tool being applied by all of these
different projects, do we need one? If so what should it do and how?
- There was a discussion of the question of how can modeling help us do better biology. A number of
questions were raised: when do you have enough information to begin modeling; when is it OK to model;
modeling can be expensive (because of the time required) and it might be cheaper to just do all the
experiments; it takes a long time to get confirmation of a model. It was emphasized, however, that modeling
is there only way one can get explicit about (even a simple) biological hypothesis.
- Can we come up with a better name for this field than Computational Morphodynamics?