Mapping lineage-traced cells across time points with moslin

Abstract

Simultaneous profiling of single-cell gene expression and lineage history holds enormous potential for studying cellular decision-making beyond simpler pseudotime-based approaches. However, it is currently unclear how lineage and gene expression information across experimental time points can be combined in destructive experiments, which is particularly challenging for in-vivo systems. Here we present moslin, a Fused Gromov-Wasserstein-based model to couple matching cellular profiles across time points. In contrast to existing methods, moslin leverages both intra-individual lineage relations and inter-individual gene expression similarity. We demonstrate on simulated and real data that moslin outperforms state-of-the-art approaches that use either one or both data modalities, even when the lineage information is noisy. On C. elegans embryonic development, we show how moslin, combined with trajectory inference methods, predicts fate probabilities and putative decision driver genes. Finally, we use moslin to delineate lineage relationships among transiently activated fibroblast states during zebrafish heart regeneration. We anticipate moslin to play a crucial role in deciphering complex state change trajectories from lineage-traced single-cell data.

CellRank 2: unified fate mapping in multiview single-cell data

“…With this formulation, we identified a putative progenitor population of medullary thymic epithelial cells missed by methods that ignore dynamics within time points. Recently, time-course studies have been combined with genetic lineage tracing [55][56][57][58] or spatial resolution [59][60][61] and emerging computational methods [21][22][23][24] use this information to map cells more faithfully across time. These enhanced inter-time point mappings can be used with our RealTimeKernel for further analysis, as demonstrated for lineage-traced Caenorhabditis elegans data in moslin 23 and spatiotemporal mouse embryogenesis data in moscot 24 .…”

“…Recently, time-course studies have been combined with genetic lineage tracing [55][56][57][58] or spatial resolution [59][60][61] and emerging computational methods [21][22][23][24] use this information to map cells more faithfully across time. These enhanced inter-time point mappings can be used with our RealTimeKernel for further analysis, as demonstrated for lineage-traced Caenorhabditis elegans data in moslin 23 and spatiotemporal mouse embryogenesis data in moscot 24 . These applications highlight the importance of our view-agnostic framework for…”

“…While we already provide multiple kernels tailored to different data modalities, current and future technologies provide additional sources of information. Concrete examples include spatially resolved time-course studies [59][60][61] and genetic lineage-tracing data [55][56][57][58] , previously already integrated in the CellRank 2 ecosystem 23,24 . Our modular interface makes CellRank 2 easily extensible toward (1) alternative single-cell data modalities by including new kernels and (2) alternative trajectory descriptions generating different hypotheses through new estimators.…”

“…The modular and robust design makes CellRank 2 a flexible framework for the probabilistic analysis of state dynamics in multiview single-cell data; it enables the rapid adaptation of computational workflows to emerging data modalities, including lineage tracing [21][22][23] We demonstrate CellRank 2's flexibility across a series of applications: using a pseudotime in a hematopoiesis context, determining developmental potentials during embryoid body formation and combining experimental time points with intra-time point information for pharyngeal endoderm development. Our approach for incorporating experimental time recovers terminal states more faithfully compared to traditional approaches mapping between time points 15 , allows studying gene expression change continuously across time and predicts putative progenitors missed by alternative approaches.…”

Unified fate mapping in multiview single-cell data

“…The modular and robust design makes CellRank 2 a flexible framework for the probabilistic analysis of state dynamics in multiview single-cell data: It enables the rapid adaptation of computational workflows to emerging data modalities, including lineage tracing 28 and spatiotemporal data 29 , the support of new data modalities with kernels, and the support of new analyses with estimators (Methods).…”