Research

The rapidly growing availability of high-throughput biological data (such as microarray, RNA-seq, exome sequence, epigenomic profiles) has opened new opportunities for addressing key questions in biology and improving health care. Success will increasingly depend on an effective computational methodology that can integrate heterogeneous data sets, effectively reduce the dimensionality of data, extract meaningful patterns out of a number of irrelevant features, and integrate across fields of biology. More and more of biological and medical sciences is becoming an information science.

My group focuses on developing such computational methods based on machine learning and statistics. Specific research projects can be divided into the following categories:

Functional genomics towards personalized health care

alt text 

more coming soon!
Personalized treatment of cancer

Collaborators: Anthony Blau (UW Medicine), Pam Becker (UW Medicine), Ray Monnat (UW Pathology), Chris Miller (UW Medicine), Fred Hutchinson Cancer Research Center, and Cardiovascular Health Research Unit

Systems biology of disease mechanisms, complex biological traits, development, evolution and tissue specificity

alt text 

More coming soon!
Cancer systems biology
Learning the regulatory networks controlling metabolic syndrome
Learning transcriptional regulatory networks driving transformation of cancer
Understanding evolution of transcriptional regulatory networks

Collaborators: Stanford Center for Cancer Systems Biology, Daniela Witten (UW Biostatistics), Jake Lusis (UCLA), Tom Drake (UCLA), Daphne Koller (Stanford), Andrew Gentles (Stanford), Maitreya Dunham (UW Genome Sciences)

Systems genetics – system-level understanding of the effect of genetic/epigenetic variation on complex traits

alt text 

Collaborators: Aimee Dudley (ISB), Daphne Koller (Stanford), Dana Pe'er (Columbia)

Genomics meets genetics – integrating functional genomic data to inform genome-wide association studies

alt text 

Pathway-based genome-wide association studies (GWAS)

Collaborators: Greg Cooper (HudsonAlpha), Sina Gharib (UW Medicine), Nona Sotoodehnia (UW Medicine)

Predictive medicine

alt text 

Collaborators: Meliha Yetisgen-Yildiz (UW Biomedical Informatics), Heather L. Evans (UW Medicine)

Developing novel machine learning and data mining techniques

alt text 

feature learning, transfer learning, sparse structural learning, feature selection

Collaborators: Daniela Witten (UW Biostatistics), Ruslan Salakhutdinov (U of Toronto), Honglak Lee (U of Michigan)

Collaborations

Our lab has close collaborations with UW Medicine, Genome Sciences, Epidemiology, UW Medical Center, Stanford Center for Cancer Systems Biology, Fred Hutchinson Cancer Research center, Allen Institute for Brain Sciences, and Seattle Children's Hospital. This enables us to access to various types of high-throuput data including genotype, microarray, RNA-seq, exome-sequencing, proteomic data and various epigenomic profiles. The short-distance collaborations also facilitate experimental or clinical validation of the hypotheses generated by our computational models, which will amplify the impact of our approaches.

Systems biology

Learning time-varying regulatory networks for understanding developmental regulation of neuron morphogenesis

In collaboration with Jay Parrish (UW Biology) and Charlie Kim (UCSF)

More coming soon!

Inferring biological processes from gene expression data

In collaboration with Serafim Batzoglou (Stanford)

alt text 
  • SUMMARY

  • Lee et al. Genom Biol 2003 [PDF]; Lee et al. NIPS 2004 [PDF]

Individual genetic variation and gene regulation

Learning the regulatory potential of sequence variations

In collaboration with Aimee Dudley (ISB), David Drubin, Pamela Silver (Harvard), Nevan Krogan (UCSF), Dana Pe'er (Columbia), Daphne Koller (Stanford)

alt text 
  • SUMMARY

  • Lee et al. PLoS Genet 2009 [PDF]; Lee et al. ICML 2007 [PDF]

Reconstructing genetic regulatory networks from eQTL data

In collaboration with Dana Pe'er (Columbia), Aimee Dudley (ISB), George Church (Harvard) , Daphne Koller (Stanford)

alt text 
  • SUMMARY

  • Lee et al. PNAS 2006 [PDF]

Sparse structure learning

Learning a meta-level prior for feature relevance from multiple related tasks

In collaboration with Vassil Chatalbashev, David Vickrey and Daphne Koller (Stanford)

alt text 
  • SUMMARY

  • Lee et al. ICML 2007 [PDF]

Efficient structure learning of Markov networks using L1-regularization

In collaboration with Varun Ganapathi, Daphne Koller (Stanford)

alt text 
  • SUMMARY

  • Lee et al. NIPS 2007 [PDF]

Efficient learning of L1 Regularized Logistic Regression

In collboration with Honglak Lee, Pieter Abbeel, Andrew Ng (Stanford)

alt text 
  • SUMMARY

  • Lee et al. AAAI 2006 [PDF]