Computational Biology Group Offers Look at Human & Fly Genome Regulation

Three new papers, co-authored by Associate Professor Manolis Kellis (head of the Computational Biology Group at CSAIL), appeared in the March 23 edition of Nature reporting three large studies of gene regulation in the human and fly genomes.

The first paper, Mapping and analysis of chromatin state dynamics in nine human cell types, offers what could be the most in-depth look thus far at the human genome outside protein-coding regions. In collaboration with Dr. Bradley Bernstein’s group at the Broad Institute and MGH, Kellis, postdoctoral fellow Jason Ernst, graduate student Pouya Kheradpour and colleagues used epigenetic information and its dynamics to reveal human regulatory networks and their role in disease.

The team used an experimental method known as ChIP-Seq (chromatin ImmunoPrecipitation followed by large-scale sequencing) to map the genome-wide locations of nine chromatin marks in nine cell types. These marks provide the cell epigenetic information that encodes the activity state of different regions of the genome. Such states encode active and poised regulatory regions (such as gene-proximal promoters and gene-distal enhancers), transcribed regions that produce RNA and repressed regions.

Kellis and his group at CSAIL developed machine-learning methods for discovering chromatin states automatically based on recurrent combinations of chromatin marks, and worked closely with Dr. Bernstein’s group to apply these methods to the nine human cell types in the context of the ENCODE project (ENCyclopedia Of DNA Elements). The resulting analysis is perhaps the most comprehensive view yet of the dynamic epigenomic landscape of human cells.

“Enhancer regions play a central role in the regulation of development, differentiation, and cell-type identity, but their connections to upstream regulators and downstream target genes have previously remained elusive because they can act at very large distances,” said Kellis.

By studying the patterns of change between distant elements and searching for pairs of elements that co-vary across the nine cell types, the researchers were able to link them into regulatory network of regulators, enhancer regions, and target genes.

“The resulting maps have far-reaching implications in the medical world and in studying the role of cis-regulatory elements in human disease,” explained Kellis.

In a handful of cases, the authors provided specific mechanisms about how a disease-associated nucleotide change was disrupting a regulatory motif responsible for an enhancer element, potentially leading to gene mis-regulation and eventually causing the disease. Tracing these networks from the nucleotides to the enhancers, regulators, and target genes can provide handles for drug targets and treatment.

“This study provides a set of tools for people working with disease and the basic biology for understanding the complexities of cells,” said Kellis. “It offers an initial glimpse into our regulatory wiring.”

Two more papers in the same issue of Nature describe Kellis’ work towards understanding the chromatin regulation and circuitry of the fruit fly genome, in the context of the model organism ENCyclopedia Of DNA Elements (modENCODE) Project.

In collaboration with the Gary Karpen and Peter Park labs, Kellis and Ernst helped detail the genome-wide chromatin landscape for the fruit fly. Their study of the fruit fly chromatin landscape provided a wealth of information on the discrete characteristics of chromosomes, genes and regulatory elements. The researchers hope to be able to grasp a better understanding of human biology through studying the functional elements of the fruit fly and how they relate to chromatin function and activity.

In collaboration with Kevin White’s group, Kellis, postdoctoral associate Chris Bristow and graduate students Pouya Kheradpour and Rachel Sealfon helped annotate the cis-regulatory map of Drosophila melanogaster, reporting their results in the same issue of Nature. Kellis and his colleagues were able to draw a map of the fly’s regulatory network across embryo development, which allowed them to infer more than 20,000 candidate regulatory elements including promoters, enhancers and insulators. The study also provided comprehensive information on transcription factor binding sites that help piece together a regulatory network of interactions between regulators and their targets.

For more on Kellis’ work and the Computational Biology group, please visit