Latent periodic process inference from single-cell RNA-seq data

Abstract

The development of a phenotype in a multicellular organism often involves multiple, simultaneously occurring biological processes. Advances in single-cell RNA-sequencing make it possible to infer latent developmental processes from the transcriptomic profiles of cells at various developmental stages. Accurate characterization is challenging however, particularly for periodic processes such as cell cycle. To address this, we develop Cyclum, an autoencoder approach identifying circular trajectories in the gene expression space. Cyclum substantially improves the accuracy and robustness of cell-cycle characterization beyond existing approaches. Applying Cyclum to removing cell-cycle effects substantially improves delineations of cell subpopulations, which is useful for establishing various cell atlases and studying tumor heterogeneity.

Machine Intelligence in Single-Cell Data Analysis: Advances and New Challenges

“…The order could be used to locate single cells along the circular cell cycle trajectory, which we called pseudo time in the cell cycle. Continuous assignment methods includes cyclum (Liang et al, 2020), cyclops (Anafi et al, 2017), peco (Hsiao et al, 2020), and Oscope (Leng et al, 2015). Cyclum and cyclops use an unsupervised learning technique autoencoder to analyze the cell-gene expression matrix.…”

“…Cyclum and cyclops use an unsupervised learning technique autoencoder to analyze the cell-gene expression matrix. To identify cell cycle phases in the scRNA-seq data, Cyclum projects high-resolution single cells onto a low-dimensional cyclic periodic trajectory, where the pseudo times are represented by radians in the range [0, 2π] (Liang et al, 2020). Unlike cyclum, cyclops uses linear projection to project data onto a closed elliptical curve in lowdimensional space (Anafi et al, 2017).…”

“…This problem also exists in studies of cell type identification, tumor development, etc. Three major methods have been developed to remove the cell cycle effect based on gene expression profiles, including Seurat (Butler et al, 2018), ccRemover (Barron and Li, 2016), and Cyclum (Liang et al, 2020). Seurat and ccRemover are the most popular ones.…”

“…Application to virtual tumor data shows that cyclum can more accurately eliminate cell cycle effects on cell clusters than Seurat and ccRemover. It can accurately distinguish two subclones in the virtual tumor data (Liang et al, 2020).…”

Melanoma Single-Cell Biology in Experimental and Clinical Settings

“…Moreover, they identified single cells with an AXL-high/MITF-low signature in an AXL-low/MITF-high population, which would have been missed in bulk sequencing and may give rise to treatment resistance. The AXL/MITF dichotomy has been supported by a later study re-analyzing these data by a new software called Cyclum to identify latent periodic developmental processes [ 81 ].…”