Chel Soumita, Gare Suman, Giri Lopamudra
Annu Int Conf IEEE Eng Med Biol Soc. 2020 Jul;2020:2425-2428. doi: 10.1109/EMBC44109.2020.9175925.
One of the major challenges in analyzing large scale intracellular calcium spiking data obtained through fluorescent imaging is to identify various patterns present in time series data. Such an analysis identifying the distinct frequency and amplitude encoding during cell-drug interaction study is expected to provide new insights into the drug action patterns over a time course. Here, we present the HDBSCAN clustering algorithm to find a clustering pattern present in calcium spiking obtained by confocal imaging of single cells. Our methodology uncovers the specific templates present in dynamic responses obtained through treatment with multiple doses of the drug. First, we attempt to visualize the clustering pattern present in time-series data corresponding to various doses of the drug. Secondly, we show that the HDBSCAN can be used for the detection of specific signatures corresponding to low and high cell density regions selected from in vitro experiments. To the best of our knowledge, this is the first attempt to optimize the clustering of calcium dynamics using HDBSCAN. Finally, we emphasize that HDBSCAN offers a high-level grasp on systems biology data, including complex spiking pattern and can be used as a visual analytic tool for drug dose selection.
通过荧光成像获得的大规模细胞内钙信号数据进行分析时，面临的主要挑战之一是识别时间序列数据中存在的各种模式。在细胞-药物相互作用研究中，这样一种能够识别不同频率和幅度编码的分析，有望在一段时间内为药物作用模式提供新的见解。在这里，我们提出了HDBSCAN聚类算法，以找到通过单细胞共聚焦成像获得的钙信号中存在的聚类模式。我们的方法揭示了通过多剂量药物处理获得的动态反应中存在的特定模板。首先，我们试图可视化与不同剂量药物相对应的时间序列数据中存在的聚类模式。其次，我们表明HDBSCAN可用于检测从体外实验中选择的低细胞密度和高细胞密度区域对应的特定特征。据我们所知，这是首次尝试使用HDBSCAN优化钙动力学的聚类。最后，我们强调HDBSCAN能对系统生物学数据有一个高层次的把握，包括复杂的尖峰模式，并且可以用作药物剂量选择的可视化分析工具。
