Nozomu Yachie

Spatial transcriptomics enables the in situ mapping of gene expression, but the validity of downstream cellular analyses relies heavily on accurate cell segmentation. Current image-based segmentation tools rely on cellular staining rather than true molecular assignment; conversely, transcript-driven tools incur prohibitive computational costs and are prone to inferring morphologically aberrant cellular boundaries.To address these limitations, TOPIC (TOpic-guided Probabilistic Image-based Cell-segmentation), a novel transcript-driven framework utilizing Latent Dirichlet Allocation (LDA), was developed. By projecting spatial RNA coordinates onto a hexagonal grid and modeling these spatial bins as documents comprising distinct gene topics, TOPIC performs a robust region-growing expansion governed by biological consistency and localized transcript density; additionally, subsequent probabilistic purification allows the model to filter out statistically unlikely transcripts to eliminate spatial cross-contamination and background noise.TOPIC was validated by comparing it against state-of-the-art tools (Xenium Multimodal, Baysor, and Proseg) across diverse tissues and disease contexts, (e.g., human breast cancer, human lung cancer, mouse brain, and mouse lung). Performance was evaluated in terms of computational scalability (global segmentation yield, overall processing speed and memory requirements), transcriptomic fidelity (cell type recovery against orthogonal scRNA-seq references, cosine similarity to reference profiles, signal sensitivity versus contamination, and spatial characterization of extracellular RNA), and morphological accuracy (2D tissue area coverage, boundary concordance with image-based ground truths, and preservation of cell-specific morphological signatures and size dynamics).TOPIC successfully overcomes image-derived geometric biases by delineating highly irregular cytoplasmic extensions (such as sprawling Alveolar Type 1 cells) that standard image-based pipelines systematically underestimate. Furthermore, TOPIC balances morphological fidelity with high biological sensitivity and signal purity; in doing so, TOPIC avoids the conservative over-segmentation frequently observed in Baysor and the unbounded acellular expansion seen in Proseg.

View record

The advent of CRISPR-Cas9 genome editing technologies has spurred the development of large-scale cell lineage tracing systems. These techniques involve integrating synthetic DNA barcodes into the chromosomes of a single cell, which are subsequently manipulated via genome editing to introduce heritable mutations. As the cell proliferates to form complex multicellular systems, there will be an accumulation of mutations as barcodes are transmitted from one cell generation to the next. At a later time of observation, single-cell RNA sequencing (scRNA-seq) can be used to capture all of the mutated barcodes in each cell, enabling the reconstruction of a tree that aims to recapitulate the cell division history. While analogous to evolutionary tree reconstruction, cell phylogeny estimation faces unique challenges owing to the scale of many organisms, as they can be composed of millions to trillions of nucleated cells. This motivated the Yachie laboratory to develop a divide-and-conquer method that aims to augment existing tools to enable large-scale reconstruction with distributed computing. A proof-of-concept of this framework was published in 2022, which culminated in the reconstruction of a tree containing over 235 million sequences. Given the promise shown by these initial demonstrations, my project focused on further developing this framework to overcome the noise, sparsity, and biases typically associated with the molecular readouts from scRNA-seq technologies. Specifically, noise and sparsity were addressed through the development of a bootstrapping strategy based on orthogonal tree agreement. A rejection sampling strategy based on sequence distance was also developed to overcome possible biases that may be encountered in real biological datasets. Together, these methods combined to provide a relatively unbiased estimate of confidence for proposed tree branches, which could be leveraged to significantly improve reconstruction accuracy.The overall findings of this project suggested widespread improvement in the rigor and robustness of the reconstruction framework, which marked meaningful steps towards accurate cell lineage reconstruction in large-scale sequencing datasets. The Yachie laboratory is currently applying the present approaches to a high-content cell lineage tracing system resolve the developmental cell division history of an adult mouse, a resource that would deepen our understanding of developmental biology.

View record

Single-cell transcriptome sequencing and cell lineage tracing technologies have enabled high- resolution mapping of different cell states and trajectories during stem cell maintenance and differentiation. Nevertheless, using such technologies, we are not able to retrieve living clones of the studied cells for further molecular or functional analysis. In this thesis, I describe the development of a technology, scCloneSelect, that enables sorting clones of a target cell state identified by single-cell RNA sequencing through combining cellular DNA barcoding and CRISPR base editing technology. Furthermore, I explore the application of scCloneSelect to study factors underlying different cell fates during stem cell differentiation, reprogramming and maintenance.

View record