Kyuri Jo

PhD Candidate
Department of Computer Science & Engineering

Seoul National University

Kyuri Jo is a PhD candidate at the Department of Computer Science and Engineering at Seoul National University. She has focused on analyzing time-series with machine learning techniques since she joined Bio and Health Informatics Lab in 2013. Her first approach to analyze biological pathway (network) with time series gene expression data successfully elucidated different mechanism of drought-resistant rice (Methods, 2014). She expanded her work to consider regulators of the dynamic network and the paper was accepted at the top conference and journal (ISMB 2016; Bioinformatics, 2016) in the fields of Bioinformatics. She has been involved in several biological research projects such as xenotransplantation and obesity researches to apply her algorithms for more effective and accurate follow-up studies. Kyuri's recent honors include a Naver Ph.D. fellowship in 2017 for her outstanding research performance in the field of computer science and the ISMB travel fellowship from ISMB 2016 for her Bioinformatics paper.

Network and Clustering Algorithms for the Analysis of Time Series Gene Expression Data

Expression levels of genes at the whole genome level can be useful for characterizing biological mechanisms. Especially, dynamic gene expression provides opportunities to understand how organisms react in specific conditions over time. However, analysis of time series gene expression data is challenging since existing methods need to be modified to consider the time dimension. In my doctoral study, I developed three new bioinformatics methods to analyze time series data in terms of network propagation. In the first study, TRAP is developed to extend the existing pathway analysis method, SPIA, for time series analysis and estimates statistical values to measure the dynamic propagation of signaling effect in the network (pathway). In the second study, TimeTP, a method to detect perturbed sub-network and regulators of the perturbation is developed. TimeTP first determines a set of perturbed sub-networks with genes connected to propagate expression changes over time by measuring cross-correlation. TimeTP extends the network to include upstream regulators of genes and evaluate the influence of regulators by influence maximization technique. In the final study, a method to infer cluster network for gene expression time series is proposed. This study aims to incorporate Gaussian process regression into the clustering process to predict unseen values of time series and provide more accurate clustering result. In addition, a network of clusters is generated by measuring similarity of expression patterns and biological functions between clusters.