Overview: Our research focuses on the development of algorithms for solving problems in molecular biology. This page provides a quick overview of the topics we are interested in, and is intended for the non-specialist. For more details, visit the links on the right panel.

Gene Regulation: The activity of genes in a cell is regulated by other genes, and a complex network of such regulatory interactions orchestrates a precise pattern of expression of genes. Such networks are often likened to circuits, such as the one here in sea urchin. By unraveling these gene regulatory networks, scientists hope to understand how a single cell develops into an adult organism, how certain diseases affect cellular function, and how new and diverse species evolve. In other words, the gene regulatory networks can unlock many of the deepest secrets of life.

We strive to unravel such regulatory networks in a variety of organisms, from yeast to honeybee, from flatworms to humans.

We build probabilistic models and algorithms that integrate the heterogeneous and noisy sources of genomics data available today, in order to deduce regulatory interactions of genes.

cis-Regulatory Evolution (CRE): The wonderful diversity of life forms we see in Nature, or in electronic snapshots of Nature, are the result of evolutionary forces, whose footprints are preserved by DNA. It is becoming increasingly clear that a majority of the morphological ("of form") diversity comes from changes in the gene regulatory network, rather than the genes themselves.

This realization drives our work in the new field of cis-Regulatory Evolution, where we use computational analysis of multiple genomes to piece together the link between gene regulation and evolution.

We work on statistical and probabilistic approaches to study how evolution tinkers with gene regulatory networks to create species diversity. In particular we use genomic data to learn how transcription factor binding sites have evolved across medium to large evolutionary distances. This subject is intimately linked with the issue of alignment (see below).

Probabilistic Alignment: The standard approach to examining the relationship among multiple sequences has been to align them using information theoretic scores. Such scores are usually not motivated by realistic models of evolution, and are instead optimized for high speed.

We are interested in developing a more realistic framework for multiple sequence comparison, that (i) models evolutionary events more explicitly, and (ii) is aware of the organization of functional elements in the sequence.

This involves development of probabilistic models, and inference on these models using genomic data, which is often a computationally challenging task.

Alignment-free sequence comparison: The future is going to be replete with genomes from a large number of species, with opportunities galore for using the genetic and molecular knowledge in one organism to hunt down the corresponding molecular players in a different organism. To be able to do this across large evolutionary distances is a highly challenging problem, because the current theory of sequence comparison relies on "sequence alignment". In the applications we are interested in, such as comparison between fruitfly and mosquito, or between honey bee and wasp, non-coding sequences are often not alignable, and novel "alignment-free" methods need development.

We are developing statistical tools for predicting functional similarity between non-coding sequences (enhancers/promoters) without relying on sequence alignment.

Alignment-free comparison of non-coding sequences can serve two purposes: (i) given the cis-regulatory modules in one species, to discover their orthologs in a highly diverged species (e.g., from fruitfly to mosquito), and (ii) given the cis-regulatory modules belonging to a pathway, to find other CRMs in this pathway in the same species.