Quantitative models of gene expression
This page is maintained as a historical record and is no longer being updated.
The Hemberg Group moved to the Evergrande Center for Immunologic Diseases in February 2021. https://evergrande.hms.harvard.edu/home
Recent technological advances have made it possible to analyze gene expression and other related properties in a high-throughput manner, and this has resulted in a wealth of data. However, the experimental data is typically large, high-dimensional and noisy. We are interested in developing computational methods that will make it possible to extract as much information as possible from the data.
Some of the ongoing research projects are:
- Inference of gene regulatory networks from single-cell RNA-seq data. Thanks to extensive annotation efforts, we have an almost complete catalogue of protein coding genes in humans and model organisms. Much less is known about how genes interact. To infer a network, one must have expression data from multiple conditions, e.g. mutants or a time-series. Due to the high levels of noise in the data and the limited number of conditions, existing methods for bulk RNA-seq have a limited ability to detect causal regulatory relations. With single-cell RNA-seq data, a more powerful approach is possible since each cell can be considered as an individual replicate. Knowing which genes interact is also key for understanding development and many diseases such as cancer and autism.
- Identification of the molecular mechanisms involved in transgenerational epigenetic inheritance. Together with the Miska lab, we are studying C. elegans to learn more about how gene expression profiles can be stably inherited. Several lines of evidence have suggested the existence of such effects, but no mechanism has been identified for endogenous genes. The short generation time and the small genome makes C. elegans a powerful model system for investigating this phenomenon.
- Identification and characterization of non-canonical secondary structures in DNA. Mutations outside of coding regions remain poorly understood and their importance in cancer and other diseases is unknown. We are investigation non-coding mutations from cancer samples to find out if the disruption of secondary DNA structures could play an important role.
- Virtual Reality technology for visualizing genomic data. We are collaborating with HammerheadVR, a leading VR development studio to develop a novel genome browser for Virtual Reality technologies.
Dr Martin Hemberg, PhD
CDF Group Leader
Martin Hemberg is a Career Development Fellow Group Leader and his research interests are centered around quantitative models of gene expression and gene regulation. He is particularly interested in stochastic models and analysis of single-cell data. Another line of research involves analyzing the role of non-coding transcripts and sequences.
Discrete Distributional Differential Expression (D3E)
D3E is a method for identifying differentially expressed genes from single-cell RNA-seq experiments. D3E compares the full distribution between two ...
Single-cell Consensus Clustering (SC3)
SC3 is a method for unsupervised clustering of single-cell RNA-seq data. In addition to a graphical user-interface, SC3 provides additional ...
Cellular Genetics Informatics
Our team provides efficient access to cutting-edge analysis methods, environments and pipelines for Cellular Genetics programme, which leads and is involved ...
Our research group uses genomics to investigate insect biodiversity and malaria transmission.
Non-coding RNA and epigenetics
We are interested in all aspects of gene regulation by non-coding RNA. Current research themes include: miRNA biology and pathology, miRNA ...
Programmes and Facilities
Open Targets is an innovative, public-private partnership that uses human genetics and genomics data at large scale for systematic drug target ...
We are interested in working in close collaboration with experimentalists as it provides with direct access to the people who generated data and have a deep understanding of the underlying biology. This approach makes it easier for us to develop mathematical models and it also provides us with a better understanding of what type of computational tools are needed.Some of the people that we have worked with in the past or are currently working with are listed below.