Computational Methods in Genomic and Proteomic Data Analysis

With the great progress of technology in genomics and proteomics generating an exponentially increasing amount of data, computational and statistical methods have become indispensable) for accurate biological conclusions. In this doctoral dissertation, we present several algorithms designed to delve large amounts of data and bolster the understanding of molecular biology. MAPK and PI3K, two signaling pathways important in cancer, are explored using gene expression profiling and machine learning. Machine learning and particularly ensembles of classifiers are studied in context of genomic and proteomic data. An approach to screen and find relations in protein mass spectrometry data is described. Also, an algorithm to handle unreliable values in data with much redundancy is presented.