Statistical and computational tools to model the structural and functional heterogeneity of skeletal muscle during regeneration

Skeletal muscle is the largest organ in the body, and is responsible for numerous critical biological functions, including locomotion, breathing, and environmental interaction. Though muscle fibers make up the majority of the skeletal muscle mass, there are many other cells that are essential to its regulation, maintenance, and regeneration. In this work, I focus on leveraging computational and statistical tools to understand the heterogeneity of cells in skeletal muscle, how its variation contributes to regeneration, and develop software to accelerate the work of future researchers through the creation of open-source programs. Finally, I explore how modern deep-learning tools can present fundamental challenges for scientific reproducibility. My first project is in using statistical analysis to estimate the clonal diversity of a population of regenerating fibro-adipogenic progenitors (FAPs). To this end, I have created a model of regeneration based on estimating a probability distribution of clonal diversity. I used a likelihood function constructed from a multinomial distribution, and used Monte Carlo simulation to create estimators of heterogeneity. In this work, I found that the clonal diversity of FAPs is compromised upon damage and progressively worsens upon repeated damage, and the effect is not explainable by ageing. My second project is to develop a novel, open-source skeletal muscle imaging software tool called FiberSight and use it to help quantify the regenerative and structural state of skeletal muscle. I built FiberSight on the widely-used open-source imaging software Fiji/ImageJ to address the challenge of analyzing diverse muscle morphologies on whole-slide tissue sections. This program features 3 major functionalities --- stain-specific Cellpose models for segmenting skeletal muscle fibers and estimating morphology, automatic quantification the central nucleation state of segmented muscle fibers, and finally robustly categorizing continuous signals of muscle fiber type based on immunofluorescence staining.