We study developmental rules and synthetically reconstruct tissue patterning
Many diseases arise from a breakdown in tissue structure, including cancers and congenital birth defects. We aim to reverse engineer the interactions between cells and their local environment that create and maintain the structure of tissues in the body.
We are particularly interested in learning to control patterning processes that occur in embryos, where the blueprint of adult tissues is established. We approach this by developing new cellular imaging, tissue construction, soft materials, and micro-scale engineering techniques.
Tissues that mimic morphogenesis by autonomously changing shape
Tissues develop by progressively increasing their complexity in shape and cellular composition. We produce synthetic tissues that mimic these processes by controlling the geometry of strains generated by cells within extracellular matrix-mimetic hydrogels. We are using these fully biological shape-shifting scaffolds to study how tissue development is coordinated during, for example, branching morphogenesis of the kidney collecting duct network. We also study how mechanics and biochemical factors such as morphogens are integrated in embryonic tissue-building processes.
SINGLE-CELL CONTRIBUTIONS TO TISSUE PATTERNING
The engineering rules that link single-cell phenotype to collective cell behavior are not well understood. Developmental processes such as cell intercalation, planar cell polarity, germ layer patterning, and body axis patterning are engineering black boxes because of this. We use cell sheet engineering, optogenetics, and single-cell analysis approaches to decode these engineering rules.
Organ-scale hierarchical patterning of organoids
Organs such as the kidney have a repetitive spatial organization of functional units that is not currently engineerable. We aim to mimic aspects of branching and folding morphogenesis to give organoid models a repeatable mesoscale structure. These efforts will make organoid systems larger and more repeatable, improving their application in drug screens and regenerative medicine.
human-computer interfaces to tackle large image set analysis
High-throughput characterization of biological image data is a bottleneck in many scientific areas. We are crowd sourcing high-quality annotations from large groups of human workers with little formal scientific training. We have developed a publicly available website, quanti.us, in close collaboration with software engineering and computer science experts. Researchers can easily setup arbitrary image annotation jobs for problems where it is difficult or expensive to produce specific segmentation algorithms. These approaches enable training of tailored machine learning algorithms. Development of human-machine image analysis pipelines will make the scientific endeavor more nimble and quantitative for life scientists.