Comprehensive Evaluation of NGS-based Structural Variant Detection

The purpose of this thesis is to explore and analyze genomic structural variations (SVs). The research includes benchmarking datasets to standardize results from various tools applied to different datasets, enabling comparative analysis.

An ensemble method incorporating consensus and machine learning approaches will be developed and tested. Visualization techniques will be employed to present the findings through various graphical representations. The project will include a web-based software platform, providing users with the tools to carry out similar analyses. The methods developed will also be applied to 550 whole genomes from the Türkiye Genome Project.

So far, we have analyzed the well-studied benchmark dataset NA12878 using 4 popular tools Manta, Delly, Lump and BreakDancer. Manta consistently shows the most true positives across different analyses, excelling particularly in capturing insertions. Analyzes are continuing with different benchmark datasets for further exploration.