Chemical-Genetic Interaction Maps


The functional and therapeutic implications of most tumor mutations are unknown. To rapidly causally link molecular alterations in tumors with drug responses we have developed methods to quantitatively map genetic interactions with small molecules. Generation of large chemical genetic interaction maps enable us to predict cancer cell line drug responses, predict new drug combinations and identify new synthetic lethal opportunities.

Previous maps we have generated identified a new synthetic lethal interaction between Dasatinib and the MYC oncogene in breast cancers (Martins et al Cancer Discovery) as well as new factors involved in PARP inhibitor resistance (Hu et al). Combined with new computational approaches we anticipate these maps can contribute to the development of new precision medicine approaches in cancer.

MYC genetic interactions
From Martins et al
Chemotherapy genetic interactions
From Hu et al

Rational Design of Combination Cancer Therapy


In cancer single agent drug treatment is rarely, if ever, curative due to drug resistance. We are developing new technologies and approaches to think about drug resistance and generate new drug combinations for use in cancers.

In one approach (Donella et al) we mapped kinase activation following drug exposure to understand how pathways are rewired by drug treatment and identify barriers to efficacy of anti-cancer drugs. This approach revealed an exciting new combinatorial approach to treat breast cancers using the combination of Aurora kinase and PI3K inhibitors. In the future we anticipate that modeling pathway dynamics and drug adaptation will be critical for the design of effective new combination therapies.

Profiling of kinase activities using chemical-proteomics
From Donnella et al
Tumor regression using combined mTOR and Aurora kinase inhibition
From Donnella et al

Interactome Guided Precision Medicine


A major barrier to our ability to make more effective treatments is our incomplete knowledge of cancer pathways and the contexts in which they are active. We develop approaches to generate new protein-protein interaction data using affinity purification followed by mass spectrometry as well as computationally analyze interactome data and integrate it with cancer patient data.

We developed a computational algorithm to integrate multi-platform -Omics data from breast cancer to identify 219 gene modules that capture the majority of molecular variation in this disease. We use this approach to delineate molecular signals coming from the tumor microenvironment and those that are preserved in cancer cell lines. These networks predict cancer drug responses and can serve as robust, high resolution biomarkers for therapies.

An approach to identify network modules in cancer and use them as biomarkers for therapeutics using Modular Analysis of Genomic Modules in Cancer (MAGNETIC). From Webber et al.