Alignment of single-cell trajectory trees with CAPITAL

Abstract

Global alignment of complex pseudotime trajectories between different single-cell RNA-seq datasets is challenging, as existing tools mainly focus on linear alignment of single-cell trajectories. Here we present CAPITAL (comparative analysis of pseudotime trajectory inference with tree alignment), a method for comparing single-cell trajectories with tree alignment whereby branching trajectories can be automatically compared. Computational tests on synthetic datasets and authentic bone marrow cells datasets indicate that CAPITAL has achieved accurate and robust alignments of trajectory trees, revealing various gene expression dynamics including gene–gene correlation conservation between different species.

Gene-level alignment of single cell trajectories

“…Furthermore, the current G2G version can only compare two linear trajectories without considering branching processes. We are aware of other efforts in aligning branched processes with DTW based tree alignment 8 . The output from such alignments, i.e., identified pairs of correspondences, could be further inputted into G2G for a comprehensive pairwise lineage alignment to capture mismatches.…”

“…Currently, dynamic time warping (DTW) is often used to align two single-cell trajectories. Several notable attempts [4][5][6][7][8][9] , such as CellAlign 5 , employ DTW to identify correspondences between two different profiles, allowing the detection of timing differences in biologically similar processes 10 . Current practice is to interpolate the gene expression time series prior to DTW, and then minimize the Euclidean distance of expression vectors between the matched time points to find an optimal alignment.…”

“…To align two single-cell trajectories, a popular dynamic programming algorithm called dynamic time warping (DTW) is often used. Several studies [5][6][7][8][9][10] including the widely-used CellAlign 6 framework employ DTW to analyze correspondences between time-course profiles, allowing detection of timing differences in biologically similar processes through the notion of warps 11 . Current practice is to interpolate the gene expression time series prior to DTW, and then minimize the Euclidean distance of expression vectors between the matched time points to find an optimal alignment.…”

Bringing CAM photosynthesis to the table: Paving the way for resilient and productive agricultural systems in a changing climate

“…By uncovering transcriptome changes at critical points of the pseudotime trajectories, the results of such analyses can not only provide crucial information about the key elements that drive cells to engage in the transition but also reveal some of the genetic components involved in the evolution from C 3 to CAM photosynthesis ( Heyduk et al., 2019 ). Similarly, the recent development of single-cell trajectory tree alignment tools ( Alpert et al., 2018 ; Conde et al., 2022 ; Sugihara et al., 2022 ) opens the door to a wide range of applications in research on facultative CAM. Comparing trajectory trees from closely related C 3 , obligate CAM, and facultative CAM species could provide answers to major biological questions about photosynthetic adaptation to drought.…”

Is your data alignable? Principled and interpretable alignability testing and integration of single-cell data

“…As a result, the methods would output some integrated cellular profiles, commonly represented as either a corrected feature matrix, or a joint low-dimensional embedding matrix, where the unwanted technical or biological variations between the datasets have been removed. These methods have played an indispensable role in current single-cell studies such as generating large-scale reference atlases of human organs [16,61,21,94,75], inferring lineage differentiation trajectories and pseudotime reconstruction [82,68,78], and multi-omic characterization of COVID-19 pathogenesis and immune response [76,2,47].…”