Gulhan Lab

Gulhan Lab

We employ statistical and machine-learning models to dissect the complexity of cancer genomes and to advance personalized oncology.

Research interests

Interpreting the patterns of mutations

Genomic instability, a hallmark of cancer, enhances tumor growth. We create algorithms to break down mutation patterns into characteristic footprints of biological processes causing genomic instability, distinguishing mechanisms such as repair deficiencies that signal vulnerabilities to targeted and immunotherapies. We leverage signature analysis and big data of cancer genomes to catalog genomic instability, using which we develop patient stratification strategies. We design tools tailored for tissue, and non-invasive liquid biopsy assays.

Mutational signatures are genomic footprints of biological processes inferred through pattern recognition algorithms. Mutation rates due to these processes vary over time and across the genome modelling these changes can reveal mechanistic insights in processes fueling cancer growth and evolution. We design comprehensive statistical models to improve the accuracy and interpretability of signature analysis techniques, aiming to facilitate its integration into clinics. 

A dynamic view of cancer

Cancer cells evolve complex trajectories and develop treatment resistance; therefore, effective profiling strategies must capture the disease's dynamic nature. 

Particularly in the very early and late stages, these evolutionary paths are not well understood. Collaborating closely with clinical researchers at the Termeer Center for Targeted Therapies and the Cancer Early Detection Clinic, and we leverage rich data resources and apply our methods to detect and study cancer at early stages and resistance in metastatic disease. The rapid autopsies and serial liquid and tissue biopsies collecged at MGH provides us an opportunity to delineate mechanisms of resistance to novel cancer treatments.

We use blood-based liquid biopsy tests for non-invasive monitoring and design machine learning algorithms to detect trace amounts of circulating tumor DNA with high sensitivity. We distinguish signals through fragment and mutational patterns amid high noise levels and inferring gene expression using epigenetic imprints on fragmentation patterns.  Additionally, by profiling precursor lesions at early stages and optimizing liquid biopsy strategies to detect circulating tumor DNA, we work towards advancing early detection and prevention strategies. 

We study the genomic and transcriptomic mechanisms that dictate the dynamics evolution and adaptation of cancer cells; particularly, long-term tumor evolution can be inferred from bulk sequencing data using mutation timing strategies and short-term cellular heterogeneity and dynamics using single-cell sequencing of cancer tissues. 

CNY 149 13th Street | Charlestown, MA 02129, USA

© Gulhan Lab