Research On Protein Spots Matching In Gel Images

With the completion of the determination of human genome sequence, a newera-post-genomic era is coming. The key task of life science has been transferred fromsequencing of genome to the functional characterization of genes. Proteomics is theimportant component of the post-genomic era, and the two-dimensional gel imageanalysis technique is one of the important techniques in proteomics. Extraction andanalysis of variant protein spots is a key content in the two-dimensional gel imageanalysis technique, and protein spot matching is the key link in variant protein spotsextracting. Protein spot matching can be used to identify potential variant proteinmolecules, and provide evidence for disease diagnosis, drug research and contaminationanalysis.By taking protein spot of the gel electrophoresis image as the research object, weexplore the algorithm of protein spot matching based on the gel’s images in this paper.The main research work, results and innovation points are as follows:Firstly, it introduced the proteomics and its key technology, explained thebackground and significance of research, reviewed the algorithm of protein spotmatching based on the gel images.Secondly, through the use of the additional landmarks and the near-neighborhoodcriteria, it presented a semi-automatic matching algorithm based on landmarks in gelimages. Region division and spot matching in corresponding region and cross-regionhave effectively raised the algorithm’s accuracy, and respectively confirm theeffectiveness of the algorithm from landmark number and image source. But thismethod was semi-automatic, needed hand work.Thirdly, based on the charactering of feature spots distribution in gel images, itpresented an automatic matching algorithm witch first feature spots match then spotsaccuracy match in gel images. At the beginning, calculated the match threshold distancewith statistics theory, then feature spots were extracted and matched based on regiondivision and similarity. after that, calculated the transformation parameter between twoimages with the matched feature spots. lastly, spots were matched accurately in gelimages. The effectiveness of the algorithm was confirmed with real gel images.Finally, for better medical effect, it presented a spots matching method witch firstpair wisely match in the group and generate a virtual gel images, then spots were matched between virtual gel images. We applied this method to the gel protein spotanalyzing software-Protein Master, and introduced all the sub-modules of spotmatching system in the Protein Master.