DataVisualization Methodologies forDataMining SystemsinBioinformatics

Bioinformatics systems benefit fromtheuseofdata mining strategies tolocate interesting andpertinent relationships within massive information. Forexample, datamining methods canascertain andsummarize thesetofgenesresponding toa certain level ofstress inanorganism. Evena cursory glance through theliterature injournals, reveals thepersistent roleof datamining inexperimental biology. Integrating datamining within thecontext ofexperimental investigations iscentral to bioinformatics software. Inthis paperwedescribe theframework ofProbabilistic Principal Surfaces, alatent variable modelwhich offers alarge variety ofappealing visualization capabilities and whichcanbesuccessfully integrated inthecontext ofmicroarray analysis. A preprocessing phaseconsisting ofa nonlinear PCA neural network whichseemstobeveryuseful todealwithnoisy andtimedependent natureofmicroarray datahasbeenadded tothis framework.