Talk 2: Future of Diverse Approaches for Multiscale Modeling (IMAG-AND Futures)

Back to IMAG-AND Futures Agenda

 

3:10-3:30 pm               Future of Diverse Approaches for Multiscale Modeling:

“Developing Context-specific hypertrophy network identifies new interactions in b-adrenergic-induced hypertrophy

Ali Khalilimeybodi, Univ. of Virginia

 

Ali Khalilimeybodi (photo)BIO: Ali earned his B.S. and Master of Science in mechanical engineering at the Sharif University of Technology, Iran in 2012. He worked in the Center of Excellence in Design, Robotics, and Automation modeling the large deformation of solid materials in the different manufacturing processes. After graduation, Ali joined the Ph.D. program at the University of Tehran and worked in the laboratory of Dr. Daneshmehr. By changing his main field of study from solid mechanics to biomechanics, Ali’s graduate work was focused on the multiscale chemo-mechanical modeling of cardiac hypertrophy, specifically on the modeling of b-adrenergic induced cardiac hypertrophy in a high blood pressure context. Ali received his Ph.D. in mechanical engineering in 2018. By pursuing the biomedical side of his Ph.D. research, he joined Dr. Saucerman’s lab at the University of Virginia as a postdoctoral research associate in January 2019. In the cardiac systems biology lab, Ali developed an automated and systematic revision method for large-scale signaling networks to derive context-specific cardiac hypertrophy networks. As a trainee, Ali aims to develop his computational and experimental abilities and prepare himself for a faculty position in the future.

ABSTRACT: Cardiac hypertrophy is a context-dependent phenomenon in which various biochemical and biomechanical factors regulate myocardial growth through a complex large-scale signaling network. Here, we developed an automated and systematic method to derive a hybrid context-dependent model for large-scale signaling networks like hypertrophy signaling, starting from its interaction graph and context-specific data. The method involves four sequential stages with an automated validation package as a core which builds an LDE model from the interaction graph and outputs the model predictions’ agreement with the experimental data. we applied a hybrid (Morris-Sobol) global sensitivity analysis in parallel with knock-out reaction test to identify the main signaling reactions in each context. By exploring all possible cases and adding new reactions between signaling components through different logic gates, we identified missing context-specific interactions and crosstalk in ISO-induced hypertrophy. We found that CaMKII could play a key role in activating non-classical pathways during ISO-induced hypertrophy. We validated the model predictions by performing immunofluorescence and western blots experiments on primary neonatal rat cardiomyocytes. This study uses different computational and experimental approaches together to identify new interactions. The diversity of the methods that you can apply for solving a biological problem in multiscale approaches was the main reason that encourages me as a mechanical engineer to study cardiac hypertrophy.

Comment

Q: From Jacob Barhak

How are you collecting data from articles? By hand? Or do you have access to some database?

Submitted by jbarhak on Tue, 03/17/2020 - 15:25

No high-quality database…

No high-quality database exists that is specific to our system, cardiac myocytes. In our experience, text-mining approaches have had a high false-positive rate. Therefore, we manually curate the literature to find appropriate experimental data. We obtain this qualitative data at steady-state or an appropriate time point from research articles with similar cell lines, assay, and experimental conditions. For each experimental data, we found at least two studies to support.

Submitted by Ali Khalilimeybodi (not verified) on Wed, 03/18/2020 - 12:42

In reply to by jbarhak

Excellent presentation. Is…

Excellent presentation. Is the calibration mainly using data measured on single time points? How do you establish confidence in the dynamical perspectives of the model behavior then?

Submitted by CZ (not verified) on Tue, 03/17/2020 - 15:31

Answer to CZ question

In the model, calibration performed on qualitative data at the model’s steady-state. While our logic-based differential equation approach does predict signaling dynamics, due to lack of sufficient kinetic experimental data we focused on steady-state conditions to study casual interactions in cardiac hypertrophy signaling network. To obtain a tuned model for each context, we calibrated the main reactions in each context by semi-quantitative data at the steady-state before applying the last module in our method to infer new interactions. In other studies, we have found that parameter estimation can be used to calibrate these models when kinetic data are available.

Add new comment