Oral Presentation 2014 International Biophysics Congress

Edge detection and segmentation of organelles, membranes and macromolecular complexes (19681)

Ben Hankamer 1 , Rubbiya Ali 1 , Michael Landsberg 1 , Rosalba Rothangel 1 , Ian Ross 1 , Christoph Gerle 2 , Henning Stahlberg 3 , Michael Hippler 4
  1. Institute for Molecular Bioscience, St. Lucia, QLD, Australia
  2. Hyogo University, Japan
  3. Biozentrum, University of Basel, Basel, Switzerland
  4. Westfälische Wilhelms-Universität, Münster, Germany

The intricate photosynthetic machinery of plants is designed to tap the huge solar energy resource of the sun (>1300 ZJ yr-1 photosynthetic reactive radiation) which is 2600x greater than our total global energy demand (the 0.5 ZJ yr-1). Consequently single celled green algae (microalgae) which can be produced on non-arable land using saline or waste water sources are being developed as advanced solar fuel systems that do not compete with food production. The importance of such systems is that they can theoretically reduce CO2 emissions, increase energy security and provide a secure basis for economic development.

The first step of all biofuel production is light capture and it's optimisation is therefore essential for the development of high-efficiency solar fuel production processes. Microalgae have approximately 20 light harvesting proteins associated with photosystems I and II which undergo complex and dynamic rearrangements in the thylakoid membranes in 4D (3D space and time). To fully understand this process requires tomographic cellular reconstructions at different time points and the ability to model macromolecular assemblies within them a ‘pseudo-atomic’ resolution. 

We have been working on methods to integrate x-ray crystallographic data into single particle reconstructions and to dock these into the contours of macromolecular assemblies within cellular tomograms. This requires accurate 3-D edge detection of macromolecules and organelles.

All the images suffer from some degree of noise contamination. Algorithms capable of high quality edge detection are therefore required and these must be able to accurately distinguish between signal and noise, and be fast as well as robust in their ability to detect objects of differing contrast, shape and size. Here algorithms developed to detect the contours of single particle macromolecules as well as macromolecules and organelles within tomograms will be presented. The performance, strengths and weaknesses of these algorithms will be summarised to illustrate the potential and limitation of using them for high throughput segmentation with the ultimate aim of generating a ‘pseudo-atomic’ resolution 3-D reconstruction of the photosynthetic machinery by docking x-ray crystallographic, single particle and tomographic data.