accepted on the recommendation of

Bioinformatics is a hybrid science at the interface between biology and computer science. For nearly two centuries, data acquisition in the wet lab and in the field has limited the pace of progress in biology. With the advent of high-throughput technologies such as genome sequencing, data analysis has become an additional bottleneck, and simultaneously a rich source of computational and statistical problems. In this thesis, we focus on two problems arising in the context of highthroughput protein identification via tandem mass spectrometry. The first problem is how to identify splice sites via protein analysis, and the second how to score the quality of tandem mass spectra. We present comprehensive, tailor-made solutions to these problems, thereby integrating theoretical analysis, algorithm design, machine learning and biological knowledge. Just a decade ago, protein identification was a fairly painstaking process which took weeks or months per protein. In the meantime, hundred