ChE PhD Dissertation Defense: Svilen Kolev

Name:
Svilen Kolev

Title:
Spatiotemporal Imaging and Quantitative Analysis of Early Gut Microbiome Assembly in C. elegans

Date:
07/29/2026

Time:
02:00:00 PM

Committee Members:
Prof. Rebecca Carrier (Advisor)
Prof. Sara Hashmi
Prof. Erel Levine
Prof. Javier Apfeld

Location:
610 EXP

Abstract:
Understanding host-associated microbiomes requires explaining not only which microbes are present, but how community states are created and lost. Entry, growth, transport, aggregation, dispersal, and clearance occur within a host environment shaped by immune, physiological, and ecological selection. Endpoint measurements collapse these processes into composition or abundance, leaving the underlying interactions and sources of inter-host heterogeneity unresolved. Caenorhabditis elegans and its model microbiome provide a tractable opportunity to observe these dynamics directly.

To observe these processes, we integrated large-scale microfluidics and programmable environmental control with hardware-synchronized fast microscopy and automated acquisition, enabling us to image gut bacterial populations in hundreds of individually confined worms for up to 20 h. The resulting heterogeneous dataset created an annotation bottleneck: gut-resident particles had to be distinguished from external bacterial signal, but exhaustive manual labeling was impractical. A data-centric workflow combining experimentally generated gut pseudolabels, task decomposition, and iterative human correction produced a reliable segmentation model for extracting longitudinal measurements of bacterial load, particle size, and position.

Alternating-label experiments followed three bacterial isolates separately across four host backgrounds. Population loads became high and broadly distributed, while single-worm trajectories fluctuated rapidly around slower trends. Within- and between-worm variation contributed substantially to total variance, supporting partial individuality rather than fixed highand low-load classes. Label-chase dynamics showed that high load did not imply stable residence: most of the load turned over rapidly, while a minority of worms retained elevated load. Particle-resolved measurements showed that small objects dominated counts and their signal cleared rapidly, whereas rare large aggregates carried disproportionate signal and were enriched in the retained tail. Aggregation did not guarantee persistence: aggregate-positive entry and later detection were distinct, microbe- and host-dependent probabilities, consistent with stochastic fragmentation, clearance, and retention.

Together, these results suggest that early assembly reflects isolate-dependent accumulation coupled to rapid turnover and stochastic aggregate-associated transitions. This process-level single-isolate baseline provides a foundation for asking how host and microbial genotypes, aging, environmental change, and additional community members reshape microbiome assembly. More broadly, it demonstrates that microbiome heterogeneity is best understood through the dynamics that generate it.

Svilen Kolev is a PhD candidate in Chemical Engineering at Northeastern University. He earned his Bachelor of Science in Chemical Engineering from the University of Massachusetts Amherst in 2019 and will defend his doctoral dissertation in August 2026. His dissertation, Spatiotemporal Imaging and Quantitative Analysis of Early Gut Microbiome Assembly in C. elegans, combines microfluidics, microscopy, machine learning, and quantitative analysis to study bacterial population dynamics inside living hosts. His work reflects a broader interest in interdisciplinary engineering and in building experimental and computational tools for quantitative biology. Svilen values collaboration and mentorship and has mentored several undergraduate researchers. Outside of research, he enjoys basketball, hiking, skiing, board games, dinner parties, and spending time with friends and family.