scREMOTE: Using multimodal single cell data to predict regulatory gene relationships and to build a computational cell reprogramming model

Abstract

Cell reprogramming offers a potential treatment to many diseases, by regenerating specialized somatic cells. Despite decades of research, discovering the transcription factors that promote cell reprogramming has largely been accomplished through trial and error, a time-consuming and costly method. A computational model for cell reprogramming, however, could guide the hypothesis formulation and experimental validation, to efficiently utilize time and resources. Current methods often cannot account for the heterogeneity observed in cell reprogramming, or they only make short-term predictions, without modelling the entire reprogramming process. Here, we present scREMOTE, a novel computational model for cell reprogramming that leverages single cell multiomics data, enabling a more holistic view of the regulatory mechanisms at cellular resolution. This is achieved by first identifying the regulatory potential of each transcription factor and gene to uncover regulatory relationships, then a regression model is built to estimate the effect of transcription factor perturbations. We show that scREMOTE successfully predicts the long-term effect of overexpressing two key transcription factors in hair follicle development by capturing higher-order gene regulations. Together, this demonstrates that integrating the multimodal processes governing gene regulation creates a more accurate model for cell reprogramming with significant potential to accelerate research in regenerative medicine.

TFvelo: gene regulation inspired RNA velocity estimation

“…Ψ g and Φ g are two sets of gene-specific, time-invariant parameters, which control the influence of TFs on the target gene, and the shape of target gene’s phase portraits, respectively. Our findings and early studies 37 have shown that it is feasible to infer the regulation relationship according to the learned parameter in linear regression models based on expression data. Additionally, linear models always have high interpretability, and rarely suffer from over-fitting.…”

“…In TFvelo, the RNA velocity , which is defined as the time derivative of RNA abundance, is determined by the abundance of involved TFs X g and itself, Using a top-down strategy, which can relax the gene dynamics to more flexible profiles 12 , TFvelo directly designs a profile function of target gene’s expression level, where Ψ g and Φ g are two sets of gene-specific time-invariant parameters, which describe the influence of TFs on the target gene, and the shape of phase portraits, respectively. Considering that linear models have been employed to represent the gene regulatory in previous studies 37-39 , is implemented with a linear model . The profile function of y g ( t )1 can be chosen flexibly from a series of second-order differentiable functions 12 , which is designed as a sine function in implementation ( Methods ).…”

AGEAS: Automated Machine Learning based Genetic Regulatory Element Extraction System

“…Moreover, CRE accessibility information combining with motif enrichment analysis is proved capable of identifying potential regulatory pathways to replace ChIP-seq based dataset utilized in the current study. 5 Refining meta-GRN reconstruction method in consideration of having input data from simultaneous scRNA-seq and single cell ATAC-seq (scATAC-seq), such as SHARE-seq published in late 2020, 63 can potentially further improve applicability of AGEAS; however, accessibility of corresponding datasets will be essential for performance assessment and method development. Similarly, initial set of classification models is also subject to change in response to future computational studies.…”

“…However, both methods may require large scale of additional background data to be applicable in other studies such as analyzing physiological or pathological development of selected cell type. The scREMOTE 5 published in 2022 is a potential method to extract generalized regulatory elements since limited background data is required with sequencing data representing sample classes. But prior knowledge on key TFs and marker genes is indispensable.…”

scDirect: key transcription factor identification for directing cell state transitions based on single-cell multi-omics data

“…to evaluate various computational methods [39,40]. Due to a large imbalance in cell types of original dataset, we subsampled equal number of each cell type in the hair follicle development system and obtained a dataset of 2,688 cells (Fig.…”