Skip to main content


Miniproject 1 - Mathematics

Supervisor: Prof Markus Kirkilionis (M dot A dot Kirkilionis at warwick dot ac dot uk)
Additional input: Adbnacer Bouchekhima (A dot Bouchekhima at warwick dot ac dot uk), Ulrich Janus (, Mike Li (M dot K dot K dot Li at warwick dot ac dot uk)
Dates: 26th March 2007 until 18th May 2007
Location: University of Warwick, Mathematics Institute, and Centre for Scientific Computing
Summary: Image analysis of Synechocystis protein distribution

Project Outline.
In this project I investigated curves of fluorescence intensity of Green Fluorescent Protein (GFP). Data came in the form of cell images of an autotrophic unicellular microorganisms, Synechocystis. The GFP had been genetically linked to a larger protein or smaller peptide sequence allowing it to cross different membrane systems and to enter different compartments of the cell. Using the intensity plots I analysed the variability of protein distributions in compartments and the membrane system of the cell. Larger samples from different cells also allowed checks on the variation of these fluorescence intensities over a whole population.

Introduction. The measurement of fluorescence intensities in unicellular organisms like Synechocystis is very challenging due to the size of the cells. Moreover the auto-fluorescence of the photosynthetic apparatus of such autotrophic microorganisms does severely influence the measurements and/or the measurement strategies. Nevertheless the simple organisation of the membrane system of Synechocystis make it an interesting species to study. In comparison to other data acquisition (mostly based on biochemistry) and microscopy techniques the advantage of confocal imaging lies in the fact that large number of samples can be measured. This holds for both an entire protein distribution in a cell, and even in addition for variations over a whole population of cells. This is important to understand the spatial organization of the cell, and how physiological differences arise.

The project was based inside Image Analysis. I analysed different cross-sections of fluorescence intensity curves over the cell, measured from a single layer dissecting the cell close to its equator. The aim ultimately being to associate the relative abundance of the protein in each of the different compartments - cytoplasm, membrane system and periplasm.

Brief Outline of work. In order to associate relative frequencies of protein abundance I began by filtering the data. Data from confocal microscopes can be quite noisy, particualrly if the gain needs to be turned up high for detection. After some consideration, a 3x3 kernal median filter was selected. Next, the images were organised, by cutting and pasting the randomly dispersed cells into a grid (Adobe Photoshop). A crucial issue in the investigation was to position the membrane system itself. With the use of the red autofluorescence channel data, and certain assumptions on the threshold, a polygon outline of the cells was produced (using Amira). Next the cell images were normalised - the principle being such that, whatever the original shape / size of the cell, post normailsation it became a precise circle. Due to the time constraints of the miniproject, the analysis of these images was greatly restricted and left incomplete.

Skills used. During the project, I learnt the basic principles of high throughput programming. More specifically I learnt how to use the Python programming language. I also learnt skills in Adobe Photoshop, and Amira.

 Further information. Please see my poster and accompanying presentation, both presented at the MOAC 2007 annual conference.