Computational methods for analyzing ngs data to discover clinically relevant mutations

Official URL: http://risc01.sabanciuniv.edu/record=b1662831 (Table of Contents)

The advent of Next Generation Sequencing platforms started a new era of genomics where affordable genome wide sequencing is available for everyone. These technologies are capable of generating huge amounts of raw sequence data creating an urgent demand for new computational analysis tools and methods. Even the simplest NGS study requires many analysis steps and each step has unique challenges and ambiguities. Efficiently processing raw NGS data and eliminating false-positive signals have become the most challenging issue in genomics. It has been shown that NGS is very effective identifying disease-causing mutations if the data is processed and interpreted properly. In this dissertation, we presented an effective whole genome/exome analysis strategy which has successfully identified novel disease-causing mutations for Cerebrofaciothoracic Dysplasia, Klippel-Feil Syndrome, Spastic Paraplegia and Northern Epilepsy. We also presented a k-mer based method for finely mapping genomic structural variations by utilizing de novo assembly and local alignment. Compared to the mapping based read extraction method, the k-mer based method improved detection of all types of structural variations, in particular detection rate of insertions increased 21%. Moreover, our method is capable of resolving complete structures of complex rearrangements which had not been accomplished before.