In Aim 2 we will construct computational models of cellular responses to drugs across genetically diverse cancer and normal cell lines. Data will be collected using a variety of single-cell and multiplex biochemical assays including sandwich immunoassays, protein and metabolite mass spectrometry, immunofluorescence microscopy and live-cell imaging of cells carrying fluorescent reporter proteins. These data will be integrated in computational models using a three-part strategy. First, significant connections between data on signaling molecules (e.g. Akt inhibition) and phenotypes (e.g. senescence v. apoptosis) will be discovered using statistical techniques such as partial least squares regression (PLSR), discriminant PLSR and Random Forest analysis. Second, network inference involving logical modeling or dynamic Bayes nets and literature-based priors will be used to determine the approximate topology of drug response networks in specific cell types. Finally, information from statistical modeling and network inference will be used to construct dynamic models in which the biochemistry of drug-target binding and of interacting response networks is rendered in mechanistic detail sufficient to reproduce and explain the observed variation in drug sensitivity and resistance from one tumor to the next. We have consciously chosen to model drug response networks for which genomic data provides clear evidence about which molecules and networks to focus on, and for which multiple precedents exist for translating cell-based studies into drug development and clinical care.
Aim 2. 1 will focus on measuring and modeling the PI3K/mTOR/Akt kinase network in triple negative breast cancer (TNBC), a disease in which this pathway is frequently mutated and being targeted by multiple kinase inhibitors in clinical development or use. Our translational goal is developing signatures and biomarkers predictive of patient response to mono and combination therapy.
Aim 2. 2 will develop new approaches to the poly-pharmacology of kinase inhibitors based on compressed algorithms that integrate diverse biochemical and structural data. We will use this information to analyze drug responses as multi-factorial perturbations of multi-component networks. Our translational goal is development of rational approaches to multi-kinase targeting.
Aim 2. 3 will focus on measuring and modeling the responses of BRAF-V600E melanoma and colon cancers to drugs such as vemurafinib with the primary aim of understanding diversity of genes implicated in acquired drug resistance. Our translational goal is overcoming or mitigating acquired resistance through design of patient-specific combination therapies using new or existing drugs.
Aim 2. 4 will compare the responses of normal and transformed cells directly with the aim of understanding the mechanistic basis of therapeutic index. We will focus on readily available "normal" human cells and on stem-cell derived cardiomyocytes, an area of interest for the FDA.

National Institute of Health (NIH)
Specialized Center (P50)
Project #
Application #
Study Section
Special Emphasis Panel (ZGM1)
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Harvard Medical School
United States
Zip Code
AlQuraishi, Mohammed; Koytiger, Grigoriy; Jenney, Anne et al. (2014) A multiscale statistical mechanical framework integrates biophysical and genomic data to assemble cancer networks. Nat Genet 46:1363-71