Research Areas

Systems biology has benefited from improvements in omics technologies, leading to an interest in the comprehensive molecular processes that bacteria go through in response to their surroundings. By specifically utilizing Chromatin immunoprecipitation and transcriptomics analysis, we can reconstruct a transcriptional regulatory network (TRN) that offers insight into gene regulation and interaction in response to external stimuli. Our interest is in understanding the transcriptional regulation in enterobacteria such as E. coli, Salmonella, Klebsiella, and methylobacterium. To conduct the high-throughput experiment data analysis, we have developed a specific pipeline and deep-learning-based high-performance software.

Genome-scale metabolic model (GSM) is computational model that integrates information about an organism's genome, metabolism and biochemical pathways to mathematically predict its metabolic phenotype under different conditions. Utilizing GSMs can provide insights into how biological system function and help in the design of microbial cell factories.

We are interested in following topics:

1. Genome-scale model reconstruction of high-value bacterial strains including Streptomyces, Vibrio, and C1 bacteria

2. Model-based analysis to understand complex biological system or to design microbial cell factories

3. Omics data integration for biological discovery

Development of computation and machine learning-based protein structure prediction algorithms allows to suggest of new proteins with improved properties and functions. Furthermore, advanced deep learning algorithms can also provide elaborate interactions between proteins and other molecules. This derived protein information can help in industrial fields and therapeutics. We are currently interested in de novo protein design, enzymatically stable protein mutation design, and optimization of interactions between proteins, ligands, and nucleic acids based on computation and machine learning.

We are currently interested in the following topics:

1. Deep-Learning based Protein-Protein and Protein-Lingand Interaction prediction

2. Protein interaction Energy Calculation based Protein Design

Antimicrobial resistance (AMR) genes pose a significant global public health threat, as bacteria increasingly develop resistance to antibiotics, leading to the spread of multidrug-resistant bacteria. To address this pressing challenge, a comprehensive understanding of the monitoring for the antibiotic resistance genes is required (e.g., genetic diversity, spatiotemporal distribution). Our laboratory is addressing this need by establishing a database of ARG information, which is aimed at facilitating not only the monitoring of ARGs, but also the development of alternative antibiotics.

The currently constructed database not only contains information on AMR genes collected from various sources, but also secured all the protein structures of all genes recorded by introducing a recently developed deep learning based 3D protein structure prediction technique (AlphaFold2, RoseTTAFold, and etc.). Through the established database, we intend to provide resources to access the latest information on the AMR gene, as well as to provide various researchers with research resources to better understand and respond to the ongoing threat caused by antibiotic resistance.

We are interested in the following topics for monitoring AMR gene:

1. Establishment, maintenance, and expansion of antibiotic resistance gene database

2. Rapid identification of AMR gene through Real Time-PCR

3. Engineering of bacteriophage enzymes with value as alternative antibiotics

4. WGS based AMR gene screening from newly isolated bacteria.

Energy storage with lithium-ion batteries have been widely used for power sources due to their high energy density and convenience for management. In response to this issue, the importance of estimating the state of health and remaining useful life for the battery system is increasing. We are currently interested in constructing general-purpose deep-learning frameworks for battery prognostics and health management. In addition, our research is not limited to energy storage, but expanded to energy generation systems such as hydrogen production. We study the machine-learning frameworks for predicting the feasibility of hydrogen production systems which are applicable to process intensified systems. 

We are interested in following topics for next-generation energy system: 

1. Macroscopic viewpoint of energy storage stability: Battery lifetime prognostics for the second life of Li-ion batteries

2. Microscopic viewpoint of energy storage stability: Battery degradation mechanisms and health of Li-ion batteries

3. Innovative hydrogen production: Optimization of hydrogen production with process intensified system

Methylobacterium is a pink-pigmented facultative methylotroph that utilizes C1 substrates as the sole carbon source. Methylorubrum extorquens (formerly Methylobacterium extorquens), a promising bacterium in methanol-based bioindustry, can be subjected to Adaptive Laboratory Evolution (ALE) to generate genetic diversity in a population of organisms and select for desirable traits. In our study, M. extorquens AM1 was subjected to ALE under higher methanol concentrations. 

Formate dehydrogenase (FDH), which catalyzes the conversion of formate to CO2 and NADH, is related to the core methylotrophy network. Among the four FDH isoforms present in M. extorquens AM1, FDH1 has been shown to possess higher catalytic activity and the smallest subunit number with only two subunits. M. extorquens PA1 meets the critical considerations for industrial strain selection, including minimal genome size, lack of plasmids, and lowest number of genes. It also shares a core methylotrophy network with M. extorquens AM1. Using M. extorquens PA1 that was knocked out three FDH isoforms, except FDH1 (M. extorquens PA1 Δfdh234), our laboratory is currently performing ALE with methanol and sodium formate. 

We are interested in the following topics to explore the potential of utilizing M. extorquens strains in industrial applications:

1. Conducting a comprehensive pan-genome analysis of all available M. extorquens strains

2. Investigating the adaptation of M. extorquens AM1 to high concentrations of methanol and identifying the underlying genes

3. Studying the adaptation of M. extorquens PA1 Δfdh234 to high concentrations of methanol and formate in order to enhance the expression of FDH1