Highlighting Differential Gene Expression between Two Condition Microarrays through Multidimensional Scaling Comparison of Lesihmania Infantum Genomic Data Similarity Matrices

摘要

Classical methods for differential gene expression between two micro-array conditions often fail to detect interesting and important differences, because these appear too little compared to the expected variability. Data fusion has proved to highlight weak differences as it allows identifying genes associated to different biological conditions. However, data fusion often leads to a new representation of data, as for example in similarity matrices. Measuring distances between similarities for each gene is not a straightforward task, and methods for this would be useful in order to find potential genes for further research. Here, we present two different kernel methods based on multidimensional scaling and principal component analysis to measure distances between genes through an example on L. infantum microarrays comparing promastigote and amastigote stages. These methods are flexible and can be applied to any organism for which microarray and other genomic data is available.