This project is created to integrate
qualitative biological knowledge for gene network reconstruction.
Reconstructing gene networks from the data produced by microarray
experiments is a hot topic in today's systems biology research. However,
most existing methods are suffered from under-sampled and noisy
microarray data. A promising direction to address this problem is to
integrate legacy biological knowledge into the gene network
reconstruction process.
Most previous works use human experts to quantify the qualitative knowledge, and/or only integrate qualitative knowledge within static probabilistic models, which greatly limits their usability. This project extends previous research works by implementing a novel approach which quantifies qualitative knowledge by using time-series microarray data, and reconstructs gene networks under the Dynamic Bayesian Networks framework. Download the pre-compiled program to try its features. This project uses the open source library on graphical models openPNL.
This is a coarse-grained protein surface visualizer. Unlike other types
of protein surface that you have seen, coarse-grained surfaces are rough and
non-continuous, and cannot be imported into popular molecular
visualization software. Besides 3-D visualization, in this project I
also used programmable vertex and fragment shaders to implement
per-pixel lighting and depth-peeling algorithm to achieve
order-independent transparency. Try a
pre-compiled executable on windows platform (You need a
graphic card supporting vertex and fragment shaders to run the
program). Collaborated with Aimin Yan.
By working on this class project I tried different machine learning
methods, including Support Vector Machine( SVM), linear regression and a
Bayesian-family method, to classify some microarray datasets in
cancer research. Take a look of
my project homepage
for details.
UBViz is a 3-D visualization tool of KEGG biological
network. Users of UBViz can download XML-format biological network file
from KEGG'sFTP site, and view the network in 3-D space with one click.
A small HTML client is integrated in UBViz so that users can open KEGG's
reference webpage for a specific component in the network.
UBViz is written in C++. It uses wxWidgets for GUI, graphviz for 3-D layout, and Expat for XML parsing.
Lucy2 is a GUI-based DNA quality trimming and vector removal tool. Lucy2
inherits all the functionalities of command-line tool Lucy(written by
Hui-Hsien Chou and Micheal Holmes). Lucy2 provides an user friendly interface
to find the unreliable region in the "raw" DNA data, so that
users can easily study the trimming parameters and results. Lucy2 is written
in C++, and built on wxWidgets.
I was working for FuelTech Inc. as an intern in the summer of 2003, my
major job was writing a distributed layer based on CORBA for Acuitiv,
their commercial 3-D CFD visualization software.
My research in VRAC was about high performance 3D CFD visualization software.
It's a combination of many technologies such as parallel computing, distributed
computing and visualization, so we call it VE-Suite. See some
Snapshots in virtual environment.
When I tried to link CORBA and OpenMP libraries in a single application, I found OmniORB and OpenMP conflict each others, so I wrote a patch which enables using OmniORB and OpenMP together.
A
keyword
searching engine for Medline documents. The system initializes with
a Medline's XML file, take user's query and return the relevant sentences
in the Medline file. Cooperated with Jing Ding and Ming Ye.
See
a snapshot and a short
movie of my computer graphics class projects.
My
first Virtual Reality project in the CprE575 class.