Ivo Sbalzarini - MOSAIC Group: Algorithms and Methods for Image-based Systems Biology

Previous and current research

The "MOdels, Simulations, and Algorithms for Interdisciplinary Computing" Group develops and applies computational methods for image-based systems biology. This includes bio-image processing, adaptive particle methods for deterministic and stochastic spatiotemporal simulations, optimization, and parallel high-performance computing. We aim at addressing significant biological challenges using novel computational methods and algorithms, without which the problem could not be solved. Our interdisciplinary group combines expertise from computer science, biology, mathematics, physics, and engineering.

Particle-method simulation of the diffusion of Green Fluorescent Protein in the 3D geometry of the Endoplasmic Reticulum (ER) of a mammalian cell during a FRAP experiment. The ER geometry has been reconstructed from confocal images using an image segmentation algorithm. The same set of computational particles (not shown) that is used to represent the geometry is also used to simulate diffusion, leading to efficient and parallel computations. The code is implemented based on the PPM Library (www.ppm-library.org).

Future prospects and goals

We develop computational methods based on the algorithmic abstraction of particles. Particle methods can cover the entire workflow of image-based systems biology from image processing to computer simulations and model/systems validation. We exploit this unifying framework to develop novel theories and tools and demonstrate them in biological applications ranging from intra-cellular transport processes to tissue development and growth. In particular, we focus on bio-image segmentation using particle methods, adaptive multi-resolution simulations using particles, and randomized algorithms for particle-based sampling and black-box optimization. All algorithms rely on efficient parallelization and implementation on multi-core computer systems using the PPM Library co-developed in our group (www.ppm-library.org).

Selected publications