Jinwook Seo

You can get more information on my bioinformatics research projects at Bioinformatics Resources in Research Center for Genetic Medicine

We generated a world largest microarray dataset under a contract from NIH NINDS to query spinal cord injury. This dataset has multiple dimensions: animal model (mouse or rat), location relative to injury site (at, above and below), severity of injury (control, mild, moderate and severe), time after injury (7-10 time points). Applying information visualization technologies such as parallel coordinates and animations, we designed and developed an interactive web interface (SpinalCordLink) to this large dataset for researchers who are tackling spinal cord injuries. There are also considerations of the analysis methods, and the entire data set was converted into three probe set algorithms (dChip, PLIER, GC-RMA), leading to nearly 10,000 microarray data files. SpinalCordLink can provide researchers with a good resource to investigate interactively one of the world largest microarray datasets.

We developed an interactive gene ontology visualization tool named GOTreePlus that can superimpose annotation information over gene ontology structures. GOTreePlus can facilitate the identification of important GO terms while visualizing them in the gene ontology structure. The interactive pie chart summary for a selected gene ontology term provides users with a succinct overview of their experimental results.

We developed an interactive visualization tool called ConSet, where users can effectively examine concordance among multiple sets at once. It uses enhanced permutation matrix visualization to enable users to easily identify similar sets. In addition to a standard Venn diagram, a new diagram (Fan diagram) was introduced, which allows users to compare two or three sets without inconsistencies.

Human clinical projects typically require a priori statistical power analyses. Towards this end, we sought to build a flexible and interactive power analysis tool for microarray studies integrated into our public domain HCE 3.5 software package. We then sought to determine if probe set algorithms or organism type strongly influenced power analysis results. We found large differences in power results based on probe set algorithm selection and noise filters. RMA provided high sensitivity for low numbers of arrays, but this came at a cost of high false positive results (24% false positive in the human project studied). Our data suggest that a priori power calculations are important for both experimental design in hypothesis testing and hypothesis generation, as well as for the selection of optimized data analysis parameters.

We hypothesized that different mRNA profiling projects have varying sources and degrees of confounding noise, and that these should alter choice of a specific probe set algorithm. We presented an interactive iterative study framework and tool for project specific probe set algorithm selection. We also showed that use of the Microarray Suite (MAS) 5.0 probe set detection p-value as a weighting function would improve the performance of all probe set algorithms.

Ph.D. research (Department of Computer Science, Human-Computer Interaction Lab). I designed and developed an interactive visualization tool, Hierarchical Clustering Explorer (HCE). There has been more than 7000 downloads from more than 60 countries since 2002. Interactive visualization techniques significantly improve users' understanding of multidimensional clustering results. I also suggested a set of orderly exploration principles (GRID principles) for an improved exploratory analysis for multidimensional data sets. The GRID principles were implemented in HCE as a rank-by-feature framework. Thousands of downloads have occurred since the public release of HCE in 2002.

I designed and implemented a PDA interface for a visitor information system using MS Embedded Visual C++ with Windows CE library.

I participated in a volume rendering research to develop a representation method and a normal vector calculation method for fast and accurate rendering of binary volume objects.

M.S. thesis topic (Department of Computer Science): I developed a web interface method for legacy OODBMS applications using Java native interface.

I surveyed and analyzed user interface requirements for Korean Enterprise Computer.