Shengyin Gu




     I am currently a Ph.D. student in the department of Computer Science, UC-Davis. I work closely with Professor Nina Amenta, Professor Bernd Hamann, Professor Patrice Koehl from UCD and scientist Inna Dubchak from Ernest Orlando Lawrence Berkeley National Laboratory (LBNL). Before I came to UCD, I received my B.S. degree for double majors in computer science and economics from Rutgers University.

     I am currently interested in computer graphics, visualization and computational geometry.

Resume:
    PDF version

Research Projects:

   Molecular Docking. (Abstract) The structure determination of protein-protein complexes remains a difficult and lengthy process, both by NMR and X-ray crystallography. Several computational methods based on docking have consequently been developed as a support and even as a possible alternative to these experimental methods. In this paper, we introduce a new protein-protein docking algorithm, shDock, based on shape complementarity. We characterize the local geometry on each protein surface with a new shape descriptor, the surface histogram. For checking the complementarity between two surface histograms, one on each protein, we use a modified Manhattan distance between their respective surface histograms. When a match is found between two patches, a model is generated for the protein complex which is then scored by checking for collision between the two proteins. We have tested our algorithm on ZDOCK (version 3.0.1) protein-protein docking benchmark. We found that for 110 out of the 124 test cases of bound docking in the benchmark, our algorithm was able to generate a model for the protein complex within a RMSD of 2.5 A from its native structure in the top 3600 candidates. For unbound docking predictions, we find a similar good model (within 2.5 A) within the top 3600 models in 54 out of 124 test cases. A comparison with other shape-based docking algorithms demonstrates the good performance of our approach.

(Left: 1AK4, a case from the benchmark, corresponds to the complex formed by human cyclophilin A (magenta) with the N-terminal domain of HIV-1 capsid (green).

1) Surface-Histogram for protein-protein docking (submitted)

2) Patricia Francis-Lyon, Shengyin Gu, Joel Hass, Nina Amenta and Patrice Koehl, Sampling the conformation of protein surface residues for flexible protein docking. BMC Bioinformatics 2010. http://www.biomedcentral.com/content/pdf/1471-2105-11-575.pdf

The ensembles can be found here.

Image not available Web History Study (Project with Google Research):

Submitted a patent application ¡°Generating Action Trails from Web History¡±.

Paper: Elin Ronby Pedersen, Karl Gyllstrom, Shengyin Gu, Peter Jin Hong, Automatic generation of research trails in web history. International Conference on Intelligent User Interfaces, 2010. http://research.google.com/pubs/archive/35610.pdf

   Geometric Representation of Proteins. (Abstract) The amino acid sequence of a protein is the key to understanding its structure and ultimately its function in the cell. This paper addresses the fundamental issue of encoding amino acid in ways that the visualization of protein sequences facilitate the decoding of its information content. We show that a featured-based representation in a three dimensional space provides an adequate representation from which the folding class of a protein as well as its domain content can be predicted.

Gu, S., Poch, O., Hamann B., Koehl, P., A Geometric Representation of of Protein Sequences, IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp 135-142, 2007. http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4413047

   TreeQ-Vista (Project with LBNL): An interactive Tree Visualization Tool with Functional Annotation Query Capabilities. (Abstract) We describe a general multiplatform exploratory tool called TreeQ-Vista, designed for presenting functional annotations in a phylogenetic context. Traits, such as phenotypic and genomic properties, are interactively queried from a user-provided relational database with a user-friendly interface which provides a set of tools for users with or without SQL knowledge. The query results are projected onto a phylogenetic tree and can be displayed in multiple color groups. A rich set of browsing, grouping and query tools are provided to facilitate trait exploration, comparison and analysis.

Gu, S., Anderson, I., Kunin, V., Cipriano, M.J., Minovitsky, S., Weber, G.H., Amenta, N., Hamann, B. and Dubchak, I.L. (2007), TreeQ-VISTA: an interactive tree visualization tool with functional annotation query capabilities, Bioinformatics 23(6), Oxford University Press, pp. 764-766. http://bioinformatics.oxfordjournals.org/cgi/content/abstract/23/6/764

   Retro-deformation of fossils. (Abstract) We consider a problem in which we want to restore the bilateral symmetry of an object which has been deformed by compression. This problem arises in paleontology, where symmetric fossils are deformed by the compression of the rocks in which they are embedded. Our input is a user-selected set of point-pairs on the deformed object, which are assumed to be mirror-images on the original symmetric object, with some added noise. We give a closed-form solution to one variant of the problem, and we show examples using this solution.

Kazhdan, M., Amenta, N., Gu, S., Wiley, D., Hamann, B., Symmetry Restoration by stretching, Canadian Conference on Computational Geometry, CCCG 2009.

Retro-deformation is a part of a broader project called Evolutionary Morphing.

Graphics and Visualization Class Projects:

­­­­­­­­ MemViz is a visualization tool of memory graph. A memory graph is a footprint of memory for a program. Each node is created for an allocated memory location and directed edges are created for memory pointers. Memory graph evolves in different stages of a running program. It can be helpful to debugging, code optimization, security improvement etc. of a program.



   Multi-class segmentation of medical data using SVM. In the world of medical image analysis, segmentation methods are important and necessary. Automatic segmentation methods outperform traditional segmentation methods which utilizes human judgment, in ways that computer methods are more consistent, reliable and many times faster; however they require human expertise at the refinement stage after automatic procedure. I implemented a modified support vector machine (SVM) and analyzed its performance for multi-class segmentation.



   First Person Shooter Game. A first person shooter (FPS) game was created by using a simple game engine I wrote. The engine is organized by scene graph. It supports quake 3 level maps, several 3D model formats, animations of characters, user control key board events, sounds effects. It implements hierarchical transformation, hierarchical bounding volumes, texturing, lighting, level of details (or LOD) and view frustum culling.



   Volume Rending. I implemented several interpolation methods in a volume render. The visual results were compared and analyzed.