Learning time-varying information flow from single-cell epithelial to mesenchymal transition data

Cellular regulatory networks are not static, but continuously reconfigure in response to stimuli via alterations in protein abundance and confirmation. However, typical computational approaches treat them as static interaction networks derived from a single time point. Here, we provide methods for learning the dynamic modulation of relationships between proteins from static single-cell data. We demonstrate our approach using TGFß induced epithelial-to-mesenchymal transition (EMT) in murine breast cancer cell line, profiled with mass cytometry. We take advantage of the asynchronous rate of transition to EMT in the data and derive a pseudotime EMT trajectory. We propose methods for visualizing and quantifying time-varying edge behavior over the trajectory, and a metric of edge dynamism to predict the effect of drug perturbations on EMT.

细胞调控网络并非静态，而是会通过蛋白质丰度与构象的改变响应外界刺激，并持续发生重构。然而，主流计算方法通常将其视为源自单一时间点的静态互作网络。本研究提供了一套方法，可从静态单细胞数据中学习蛋白质间相互作用的动态调控机制。我们以转化生长因子β（transforming growth factor beta, TGFβ）诱导的小鼠乳腺癌细胞系上皮间质转化（epithelial-to-mesenchymal transition, EMT）为模型，利用质谱流式细胞术（mass cytometry）进行样本表征，对所提方法进行了验证。我们利用数据中EMT进程的异步性，推导得到了EMT伪时间轨迹。我们提出了可在该轨迹上可视化并量化动态变化的互作边行为的方法，同时构建了一种互作边动态性指标，用于预测药物扰动对EMT进程的影响。