%0 Conference Proceedings
%4 sid.inpe.br/sibgrapi@80/2008/07.19.00.24
%2 sid.inpe.br/sibgrapi@80/2008/07.19.00.24.13
%@doi 10.1109/SIBGRAPI.2008.42
%T Exploratory visualization of RFLP-PCR genomic data using Multidimensional Scaling
%D 2008
%A Llerena, Soledad espezua,
%A Maciel, Carlos Dias,
%@affiliation student
%@affiliation professor
%E Jung, Cláudio Rosito,
%E Walter, Marcelo,
%B Brazilian Symposium on Computer Graphics and Image Processing, 21 (SIBGRAPI)
%C Campo Grande, MS, Brazil
%8 12-15 Oct. 2008
%I IEEE Computer Society
%J Los Alamitos
%S Proceedings
%K data visualization, multidimensional scaling, LMDS, dimensional reduction, RFLP-PCR.
%X Gel electrophoresis images are common results of biomolecular techniques such as RFLP-PCR (Restriction Fragment Length Polymorphism - Polymerase Chain Reaction). These images are used to discover the genetic relations between organisms. Find patterns in these images are a complex and delayed work if it is performed by humans. Traditionally, the analysis of gel electrophoresis images has been done by biologist using dendrogram representations aiming to capture the relations between organisms in a hierarchical organization. However this representation may be hard to analyze, especially when the information becomes large. This highlights the need to seek new ways of representing this type of data that become more intuitive. One of that methods is MDS (Multidimensional Scaling) which is a method used to transform measurements of similarity (or dissimilarity) between pairs of objects into points in a low-dimensional space, allowing visualize the data in a form that makes it easier to interpret. This paper proposes a new procedure to represent RFLP-PCR images as points in a low-dimensional space, based in a MDS technique. The procedure was applied in a genomic dataset obtained from a Brazilian collection of N2-fixing bacterial strains belonging to the genus Bradyrhizobium. The results showed the efficacy of the procedure to represent the RFLP-PCR images, facilitating the identification of patterns in a more intuitive way than dendrogram's representations. Also, our procedure allows an appropriate integration with a pattern-recognition algorithm, taking advantages of the visual human skills and the computational power.
%@language en
%3 espezua-MDS.pdf