Continuous-state HMMs for modeling time-series single-cell RNA-Seq data

Abstract

Motivation Methods for reconstructing developmental trajectories from time-series single-cell RNA-Seq (scRNA-Seq) data can be largely divided into two categories. The first, often referred to as pseudotime ordering methods are deterministic and rely on dimensionality reduction followed by an ordering step. The second learns a probabilistic branching model to represent the developmental process. While both types have been successful, each suffers from shortcomings that can impact their accuracy. Results We developed a new method based on continuous-state HMMs (CSHMMs) for representing and modeling time-series scRNA-Seq data. We define the CSHMM model and provide efficient learning and inference algorithms which allow the method to determine both the structure of the branching process and the assignment of cells to these branches. Analyzing several developmental single-cell datasets, we show that the CSHMM method accurately infers branching topology and correctly and continuously assign cells to paths, improving upon prior methods proposed for this task. Analysis of genes based on the continuous cell assignment identifies known and novel markers for different cell types. Availability and implementation Software and Supporting website: www.andrew.cmu.edu/user/chiehl1/CSHMM/ Supplementary information Supplementary data are available at Bioinformatics online.

Inferring TF activation order in time series scRNA-Seq studies

“…While HMMs have a (finite) well defined set of states, CSHMM can have infinitely many states (which we use to represent continuous time of cells). CSHMM-TF extends the formulation of CSHMM for time-series scRNA-Seq data (first presented in [21]) by adding TF regulation information to each path (edge). In addition, the model also assigns the time at which a TF is impacting its targets.…”

“…For model initialization, the advantages of the SCDIFF initialization method [15] for CSHMMs have been previously discussed in [21]. Based on these results we use the same initialization for CSHMM-TF as well.…”

“…However, unlike traditional HMMs they have infinitely many states and so can be used to represent continuous time. The continuous states are used to determine assignment of cells to paths in the model (Fig 1) and transition probabilities are used to denote branching of cells to different fates [21]. Here we extend this model to take into account TFgene interaction as well.…”

Single-cell time-series mapping of cell fate trajectories reveals an expanded developmental potential for human PSC-derived distal lung progenitors

“…The new method relies on Continuous State Hidden Markov Models (CSHMM). CSHMMs allows us to combine the continuous representation offered by current dimensionality reduction methods with the ability to handle noise, dropouts and identify regulators based on its probabilistic assumptions (Lin and Bar-Joseph, 2019).…”

“…We used Continuous State Hidden Markov Models (Lin and Bar-Joseph, 2019) to reconstruct the branching process of the data. The model was first initialized by clustering cells at each time point.…”