Novel multimodal approaches of decoding cellular dynamics for effective cell fate manipulation

背景：单细胞技术的最新进展为去卷积细胞异质性和研究细胞分化过程中每个细胞群体的细胞动力学提供了新的机会，但现有的单细胞方法大多集中在单一模式上。尽管这些单峰方法对细胞状态动力学产生了重大的变革性见解，但它们受到细胞状态描述不完整的限制，从而阻碍了更有效的细胞命运操纵的发现。不断增加的单细胞多组学数据的可用性提供了前所未有的机会来克服这一限制，从而获得对细胞分化中的细胞动力学的深刻理解。然而，从单细胞多组学数据中破译细胞分化过程中的细胞动力学所需的多模式方法仍然非常缺乏。目标：我的长期职业目标是用机器学习的计算方法破译各种生物过程中复杂的细胞动力学(细胞状态如何在时间和空间上变化)。由于大多数生物过程中的细胞状态本质上是异质性的，因此单细胞测量，特别是能够全面描述细胞状态的单细胞多组学数据，对于破译细胞动力学是必不可少的。因此，作为未来五年的中期目标，我将专注于开发多模式方法，从单细胞多组学数据重建细胞动力学和细胞分化的潜在调控网络。这一中期目标由三个相辅相成的短期目标组成目的1：开发一种多模式聚类和注释方法，从单细胞多组学数据中识别细胞分化中的细胞群体。目的：从单细胞多组学数据中为细胞分化中已识别的细胞群体重建细胞动力学和潜在的基因调控网络。目的：根据单细胞多组学数据重建细胞分化过程中的细胞动力学，建立模拟细胞命运操纵的计算机内模型。影响：单细胞组学数据不会自动导致在细胞分化中驱动更有效的细胞命运操纵的预测。大多数现有的计算单细胞方法是单峰的，因此对所研究的生物过程只提供了最低限度的了解，这阻碍了从单细胞数据集进行生物学发现。单细胞多模式方法将弥合这一差距，从而极大地提高我们对细胞动力学的理解。我们开发的计算重建的细胞动力学模型可以用来探索在各种应用中操纵细胞命运的新策略，如细胞和组织工程，以及生物燃料的生物合成，所有这些都具有巨大的社会经济效益。

BACKGROUND: The recent advance of single-cell technologies presents new opportunities to deconvolve cell heterogeneity and examine the cellular dynamics for each cell population in cell differentiation, yet existing single-cell methods mostly focus on a single modality. Even though these unimodal methods have yielded significant transformative insights into cellular state dynamics, they are limited by the incomplete profiling of the cellular states and thus impede the discovery of more efficient cell fate manipulation. The ever-increasing availability of single-cell multi-omics data offers unprecedented opportunities to overcome this limitation and thereby derive a deep understanding of cellular dynamics in cell differentiation. However, the multimodal methods required to decode cellular dynamics in cell differentiation from single-cell multi-omics data are still greatly lacking. OBJECTIVES: My long-term career goal is to decode the complex cellular dynamics (how the cellular states change temporally and spatially) in various biological processes with computational approaches employing machine learning. Since the cellular states in most biological processes are substantially heterogeneous, single-cell measurements, particularly single-cell multi-omics data that could comprehensively profile the cellular states, are indispensable for decoding the cellular dynamics. Therefore, as a mid-term objective in the next five years, I will focus on developing multimodal approaches to reconstruct the cellular dynamics and the underlying regulatory networks in cell differentiation from single-cell multi-omics data. This mid-term objective is composed of three complementary short-term aims Aim1: Develop a multimodal clustering and annotation method to identify cell populations in cell differentiation from single-cell multi-omics data. Aim2: Reconstruct the cellular dynamics and underlying gene regulatory networks for identified cell populations in cell differentiation from single-cell multi-omics data. Aim3: Build an in-silico model to evaluate simulated cell fate manipulation based on reconstructed cellular dynamics in cell differentiation from single-cell multi-omics data. IMPACT: Single-cell omics data do not automatically lead to predictions that drive more effective cell fate manipulation in cell differentiation. Most existing computational single-cell approaches are unimodal and thus provide only a minimal understanding of the studied biological process, which impedes biological discoveries from the single-cell datasets. Single-cell multimodal approaches will bridge this gap and thus dramatically improve our understanding of cellular dynamics. The computationally reconstructed cellular dynamics models that we develop can be used to explore new strategies to manipulate cell fates in various applications such as cell and tissue engineering, and the biosynthesis of biofuels, all of which have great socio-economic benefits.