September 2005 - May 2012
The Myers Lab is developing algorithms and software for the automatic interpretation of images produced by light and electron microscopy of stained samples, with an emphasis on building 3D and 4D "atlases" of brains, developing organisms, and cellular processes.
The central question that we are interested in is how do all the molecular agents encoded in a genome create an organism. How from a single, maternally patterned zygote, does the genome, with incredible reproducibility, direct the growth and differentiation of cells to produce an individual? As computer scientists, our aim is to produce platforms, algorithms, and software so that together with domain experts, we can accelerate the pace at which the intricate answers to this basic question are discovered.
While biologists have been interested in development and anatomy for a very long time, three recent shifts have transformed the experimental landscape: (1) the sequencing of all the common model organisms, (2) rapid advances in light microscopy, and (3) an expanding panoply of genetically encoded reagents. Together these open up the possibility that we can map anatomy and developmental trajectories in terms of molecular agents on a comprehensive scale. Given such data sets, we further believe it may be possible to begin to make real breakthroughs on how the genomic program creates specific shapes and patterns reproducibly.
While there are significant obstacles to be overcome in terms of imaging technology and molecular reagents, the most significant bottleneck in this vision is the existence of robust and reliable algorithms and software for interpreting the information that is present in the tens of thousands of 3D stacks and time-series movies that are produced by such projects. The Myers lab began to focus on this informatics specialty after Gene left Celera in 2002, based on the belief that it holds the greatest potential for big breakthroughs in molecular and cellular biology. Today our entire research focus is along these lines and we have even begun to foray into the development of microscopes and robotics to realize high-throughput pipelines.
The lab is highly collaborative and works with many investigators on computation problems for specific investigations. The list below gives only the broad direction and themes for the group.
A Light-Based Map of Every Neuron in the Fly Brain
Fly brains, after deformable registration to the pattern of neuropil, are stereotypic to within 1-2 microns. By using promotor driver lines and Cre-recombinase constructs, we plan to capture 3D stacks of 100,000 randomly sampled neurons from the fly core brain that contains about 20,000 neurons. In effect, we are performing a 5X shotgun sampling of the neurons in the fly brain, and with our algorithms and software expect to be able to provide a model of almost every neuron in the brain along with information about variance in the structure of the brain.
Perfect Cell Lineage Tracking Through the First 24 Hours of Development of a Fly or Zebra Fish
Next generation SPIM microscopy portends a sampling rate and resolution that may make it possible, with sufficiently sophisticated software, to trace the lineage of every single cell in a developing embryo for about 24 hours. For a fly, this would imply one could monitor development up to the point where the embryo is about to hatch and become a larva. With such a platform in hand, one can then imagine an incredible variety of markers (e.g. every transcription factor) that one might like to monitor through this important developmental arc.
A Comprehensive Library For In-Vivo Monitoring of Intra-Cellular Processes
Monitoring intracellular processes is difficult because of the scale involved. Nonetheless, we have had significant success monitoring things such as centrosome formation, microtubule growth, nuclear envelope breakdown, etc. We continue to solve these kinds of problems on a case-by-case basis with our collaborators, but I still imagine a project in which a set of systematic and comprehensive assays for various cellular functions are realized and widely employed to gain a deeper understanding of the biophysical and signaling processes within a cell.
Diffraction-Limited Imaging of a Mouse Brain Volume in a Day
We have built a multi-photon microscope that is 60 times faster than a conventional multi-photon and have begun engineering an onboard microtome that will enable us to image the entire volume of a mouse brain in six days with no supervision after the initial set up. A next-generation version that we are contemplating may do this in a day. We are currently exploring with collaborators the important questions of (a) how to best prepare samples histologically and (b) what range of experiments one can perform with this capability.