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.

Reconstructed Single-Cell Fate Trajectories Define Lineage Plasticity Windows during Differentiation of Human PSC-Derived Distal Lung Progenitors

“…For this, we extended our previously developed computational method based on Hidden Markov models (HMMs) (Ding et al, 2018;Rashid et al, 2017) in order to continuously assign cells along trajectories while still being able to infer regulators controlling branching events, hereafter referred to as a CSHMM (see STAR Methods). This model allowed us to combine the continuous representation offered by current dimensionality reduction methods with the ability to handle noise and dropouts and identify regulators based on its probabilistic assumptions (Lin and Bar-Joseph, 2019). Unlike standard HMMs, which are defined using a discrete set of states, CSHMMs can have infinitely many states allowing for continuous assignment of cells along developmental trajectories.…”

“…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.…”

“…In other words, large distance from the line (either up or down depending on the number of over or under expressed genes) can be used to observe an earlier split by noticing that cells start to move away from the line at a certain point (for example, P7 cells are fairly noisy and not as well represented by the model as P1 cells and as a result are further from the line). Further details are available in: (Lin and Bar-Joseph, 2019).…”

Inferring cell-cell interactions from pseudotime ordering of scRNA-Seq data

“…We start by using a trajectory inference method to obtain grouping and pseudotime ordering for cells in the dataset. Here we use continuous state Hidden Markov model (CSHMM) [10] for this, though as discussed below, TraSig can be applied to results from other pseudotime ordering methods. We then reconstruct expression profiles for genes along each of the edges using sliding windows summaries.…”

Optimal transport analysis reveals trajectories in steady-state systems

“…A large collection of trajectory inference methods have been developed in recent years [38] to address this issue. These methods broadly fall into two classes [20]: (1) methods that deal with a single stationary snapshot observed from a cellular population at equilibrium [42, 37, 31], and (2) methods that deal with a time series of snapshots from an evolving population [29, 36, 20].…”