Computer analysis of electrophoresis gels images for the molecular characterization of organisms

Molecular characterization is the process in which conspicuous attributes of specific molecules are determined. This work focuses on the development for algorithms to be used in a computational system to support tasks of molecular characterization of organisms, which allows to link particular features to organism according to the composition of molecules such as DNA, RNA, proteins, etc. Such composition is evaluated in this particular case through electrophoresis gels, whose images usually exhibit low contrast and distortions that draw their use difficult.

This project contributes with the creation of a computational tool that improves the electrophoresis gel images utility, by reducing the invested times in their analysis and by augmenting the reliability and robustness of the obtainable results,, compensating errors originated in human factors, and making easier the recovery of hidden information. Additionally, the tool is capable of managing metainformation associated to the images, including the whole process from the sample taking to the image capture, with the goal to allow advance search of data based on that metainformation.

This project is now three years young, which is approximately one third of the time of existence of similar projects in Ireland and Austria that have been publishing their results. Nonetheless, this report shows valuable contributions in the proposed algorithms, capable to improve the degree of process automation in the molecular biology laboratories

The main results reside, on one side, in the consolidation of the proposed architecture for the whole systems, as well as the prototype implementation of the capture system, the compensation of optical and perspective distorsions, a first proposal for the lane detection, the evaluation and selection of background removal algorithms and a novel strategy for the smile effect correction, based on active shape models.

This project is continued in the new research activity entitled “Computer analysis of electrophoresis gel images: advanced methods for the meta-information management and digital image processing” managed by to the Electronics Engineering School of the Costa Rica Institute of Technology.

Keywords: image analysis, lane segmentation, background removal, smile effect, PCR, electrophoresis, molecular characterization of organisms, databases