Angiogenesis in Health and Disease: In-vivo and in-silico
PI: Petros Koumoutsakos (ETH Zurich)
Co-PIs: Luisa Iruela-Arispe, Mauro Pezzè, Igor Pivkin
January 1, 2014 - December 31, 2016
Angiogenesis, the process of generating new vessels from pre-existing vasculature, is essential for the growth, repair and function of all vertebrate tissues and a hallmark of cancer. The mechanisms that regulate the process of angiogenesis have been greatly deconstructed at the molecular and signaling level, however, although acknowledged, the contribution of physical forces, including essential flow related phenomena, are poorly understood. More importantly, the integration of molecular and physical information, on the predicted vascular pattern and structure is inefficient when performed experimentally one step at the time. We propose an integrative computational/experimental study to investigate the role of multi-scale flows in angiogenesis through High Performance Computing. The success of this project relies on the capability of mapping effectively multiscale numerical methods to efficient software of massively parallel computer architectures so as to access the multitude of spatiotemporal scales inherent to angiogenesis. The integration of simulations with in vitro experiments will be performed through a Bayesian uncertainty quantification framework. The present computational+experimental approach aims to accelerate scientific discovery and to elucidate the role of physical forces in angiogenesis and it is envisioned that it will have an impact on regenerative medicine. We aim to develop software that will find broad utility for scientists in Computational Life Sciences, while making fundamental contributions that can be incorporated in other projects in the PASC network. We distinguish domain/project specific and broader computational /software challenges and proposed actions.
Domain/Project specific challenges
Current simulation capabilities in angiogenesis (as well as other areas of Life Sciences) are challenged by the complex unsteady geometries and the interaction of diverse physical phenomena across scales. We propose the development of a multi-scale particle-mesh framework for simulations in conditions relevant to in-vitro experiments and in-vivo angiogenesis.
We recognize the importance of bridging in-vivo/in-vitro experimental data and computational models of angiogenesis.
We propose the development of a Bayesian uncertainty quantification framework to quantify parameters of computational models using data from relevant experimental studies. In turn we note that the robustness, accuracy and the efficient hardware implementation of computational methods is of paramount importance as they will be embedded in uncertainty quantification algorithms that require multiple realizations.
The problem of integrating grid and particle based models, that describe physical systems across different scales, is fundamental in computational science. We plan to build a general and reusable infrastructure consisting of a set of reusable software components supported by a general purpose integration mechanism. We plan to create the angiogenesis toolset on top of this infrastructure to facilitate the proposed uncertainty quantification studies. We plan to distribute the infrastructure suitably documented to the computational science community so that it can be reused by other projects, such as in Astrophysics and Fluid Dynamics.
The proposed angiogenesis toolset is a powerful tool for biologists and physicians for clinical applications. We plan to develop both technical and general-purpose interfaces. The technical interfaces will be designed for computational science specialists and will allow access to all the technical details, the general-purpose interfaces will be designed for biologists and physicians and will allow easy access to the tool by specialists of clinical experimentation who need quick and intuitive access to the features needed for the experiments.
In order to accomplish these goals we will develop novel technology that integrates modeling of multiple physical processes to enable the prediction of vascular patterns and the quantification of the associated processes during angiogenesis. A hierarchy of multi-scale computational tools will account for chemical signaling, the detailed structure of blood cells and their transport in the growing vascular network. Experiments and imaging techniques will guide the development of the three-dimensional computational models and will provide data necessary for the quantification of uncertainty in their parameters. We will introduce a Bayesian uncertainty quantification framework in order to integrate the information available from in-vivo and in-vitro experiments with computational models. This framework will be used in turn to guide further in-vitro and in-vivo experiments so as to enhance in a systematic way the predictive capabilities of the computational models. We consider that our approach can establish a paradigm shift in the way in-vivo, in-vitro and in-silico studies are integrated demonstrating advances in Life Science as enabled by HPC.
We note that an important contribution of the project is the design and development of a reusable software infrastructure to integrate grid based and particle based models. This infrastructure will provide a general purpose integration framework to be reused by other computational scientists to solve problems that require the integration of heterogenous components, similar to those encountered in this project for angiogenesis. Building a general purpose integration framework, that effectively supports the efficient and seamless integration of heterogeneous components and techniques of computational science, is an open challenge of software engineering. Indeed, to be effective, such framework should provide interfaces and communication mechanisms general enough to easily integrate disparate and technologically different components and, at the same time, the framework interfaces and mechanisms should be also specific enough to allow an efficient communication among the integrated entities. As a consequence, designing and developing this framework requires a careful investigation of its potential applications and a careful evaluation of the trade offs between generality and effectiveness of the framework interfaces. We will instantiate the reusable infrastructure to integrates a flow simulations engine together with a molecular dynamics engine and we will design different interfaces for computational scientists and biologists to allow the smooth integration of the angiogenesis tool suite in clinical practice.