IGSP Home » Labs » Furey » Projects

Research & Collaborations

Charomatin Structure and Epigenomics

Open Chromatin in Human
Greg Crawford (IGSP), Jason Lieb (UNC), Vishy Iyer (UT-Austin), Ewan Birney lab

Data from genome-wide DNaseI hypersensitivity experiments (Crawford) and Formaldehyde Assisted Isolation of Regulatory Elements (FAIRE - Lieb) experiments identify regions of open chromatin in the genome where gene regulation elements such as transcription factors, enchancers, repressors, and insulators bind in a particular cell type. We are developing computational methods to combine these data to produce an open chromatin map of the human genome (Birney). ChIP experiments (Iyer) on those same cell lines provide an initial functional annotation of many sites of open chromatin. As part of the ENCODE Consortium, we will be generating and analyzing data from 15-20 cell lines over the next four years to identify and better understand functional regions in the human genome.

Related Publications:

X-inactivation
Hunt Willard (IGSP), Zhong Wang (IGSP), Sayan Mukherjee(IGSP)

A significant portion of genes on the inactivated human X chromosome has been found to be transcriptionally active. We have investigated the role of DNA sequence in this process. Using sequence feature, we have created accurate classifiers that can distinguish genes that escape inactivation from those that are subject to inactivation.

Related Publications:

Early Replicating DNA
David Kaufman (UNC-Chapl Hill), Stephanie Cohen (UNC-Chapel Hill), Norman Doggett (LANL)

We have identified regions of the genome that replicate early in S phase in normal human fibroblasts. These regions have been found to have distinct genomic profiles when compared to other later replicating regions. Currently, we are developing a classifier to predict early replicating regions. We also aim to explore whether the replication timing of these regions has been evolutionarily conserved.

Related Publications:

  • Cohen SM*, Furey TS*, Doggett NA, Kaufman DG.
    Genome-wide sequence and functional analysis of early replicating DNA in normal human fibroblasts.
    BMC Genomics. 2006 Nov 29;7:301. (*- Contributed equally)
    PMID: 17134498 [PubMed - indexed for MEDLINE]

Data:

Cancer Genomics and Disease Research

Integration of heterogeneous cancer data
Sayan Mukherjee(IGSP), Phil Febbo(IGSP)

The goal of our lab is to develop statistical methods and computational software that integrate high-dimensional heterogeneous but complementary data from cancer samples to identify fundamental genomic alterations with clinical and biological relevance. Analyses are performed in the space of gene sets, biologically related sets of genes such as those belonging to a signalling pathway, instead of at the single gene level. Classification models are constructed using the multi-task learning algorithm to allow the simultaneous consideration of several types of data while constructing the classifiers. Unlike other methods, these models are readily interpretable biologically and identify gene sets enriched in a single data set or across data sets. Specific analyses being performed are investgating sensitivity to drugs using expression and copy number data.

Disease Association Studies
Beth Hauser (Center for Human Genetics, Duke University)

Association studies involve genotyping populations using a number of polymorphic markers, usually single nucleotide polymorphisms (SNPs). Correlations between certain polymorphisms a disease phenotype are then used to identify regions of the genome that may be involved in that disease. Generally, these regions of interest are large (>1Mb). We are investigating the use of genome sequence features in the analysis of association study data in order to help better identify candidate genes or regulatory regions within these large regions of association.

Related Publications:

  • Wang T, Furey TS, Connelly JJ, Shihao J, Nelson S, Heber S, Gregory SG, Hauser ER.
    A novel genome-scale transcription factor binding site prediction method and its application to candidate gene identification in human disease
    Under review.

DNA Hypermethylation in Cancer
Susan Murphy (IGSP)

In normal cells, DNA methylation contributes to gene regulation by epigenetically silencing expression of certain genes. In cancer cells, regions of the genome become aberrantly hypermethylated causing the silencing of key genes that may contribute to the cancer phenotype. We are investigating the relationship between DNA in regions around the transcription start site of genes and their propensity of becoming hypermethylated in cancer cells. Using these features, we are predicting novel genes that are prone to hypermethylation. Predictions are being tested in the Murphy lab in primary ovarian cancers.

Related Publications: