Jun Huan
Location: (Lawrence, KS )
Personal Research Web Page: http://people.eecs.ku.edu/~jhuan/
Keywords: Data mining and machine learning: theory, algorithms, & applications, Bioinformatics and Biomedical Informatics: functional genomics, chemical genomics, metabolomics, & systems biology
Posted on: Monday, June 1st, 2009
Broad Research Area: AI / Machine Learning / Robotics / Vision, Databases / Information Retrieval / Data Mining, Scientific/Medical Informatics
Research Interests:
My research interest is to develop and apply data mining and machine learning algorithms to accelerate knowledge discovery in science and engineering disciplines including medicine. Our current focus is on exploring biological systems at two levels. At the molecular level, we focus on mapping out biomolecule interactions at the whole genome level and identifying the connections between biomolecule interactions and clinic endpoints such as disease diagnostics and personalized medicine development. At the system level, we focus on identifying the dynamic control mechanisms of complex systems to improve the modeling and engineering of biological systems. In our investigations we rely on high-throughput and low-throughput experimental data. These data include protein sequences and structures, chemical structure-activity profiles, gene expression profiles, protein-ligand interactions, biological pathways, DNA copy number variations, single nucleotide polymorphism (SNP), chemical toxicity, and disease phenotypes. The applications of our work could be found in Immunology, Neurology, Toxicology, and Drug design, such as:
* Protein functional annotation, including enzymatic functions and protein-ligand interactions prediction
* Fold recognition of protein sequences, especially pathological genes
* Leads optimization and ADME-Tox prediction in drug screening
* Chemical probe identification
* Biological system identification, including disease pathway and chemical toxicity pathway reconstruction
Though the problem set is diverse, the common threads of our work are geometric and probabilistic representations of biomolecules and their interactions, sparse pattern search in vector and kernel spaces, and learning generative and discriminative models with regularization. Much of our work addresses two core problems in data mining: stable pattern identification and mining structured data in Non-Euclidian spaces.
