Thema der Dissertation:
In search of the mirror image: In silico prediction and similarity scoring of tandem mass spectra in metabolomics and metaproteomics Thema der Disputation:
Computational Strategies for Compound and Peptide Identification from Tandem Mass Spectra
In search of the mirror image: In silico prediction and similarity scoring of tandem mass spectra in metabolomics and metaproteomics Thema der Disputation:
Computational Strategies for Compound and Peptide Identification from Tandem Mass Spectra
Abstract: Accurate identification of compounds and peptides in biological samples is fundamental to advancing our understanding of biological systems, with broad applications in biology and medicine. Tandem mass spectrometry enables molecular characterization through the analysis of fragmentation patterns. However, the scale and complexity of modern high-throughput experiments necessitate computational methods capable of efficiently and accurately interpreting these data.
The first talk provides an overview of computational strategies for the identification of compounds and peptides from tandem mass spectra. It introduces approaches including spectral library search, protein database search, molecular fingerprinting, in silico spectral prediction, and de novo spectral annotation, covering applications in both proteomics and metabolomics. Key algorithms that underpin these approaches are discussed.
The second talk focuses on in silico spectral prediction and examines its potential and limitations for molecular identification. Mistle, a high-speed spectral search engine, is presented and evaluated for its ability to improve identification rates in large metaproteomics search spaces by searching complete in silico spectral libraries. Subsequently, FIORA, a novel graph neural network-based fragmentation algorithm, is introduced for in silico prediction of small-molecule metabolite spectra. Together, our findings demonstrate the potential of spectral prediction to aid computational identification of compounds and peptides.
The first talk provides an overview of computational strategies for the identification of compounds and peptides from tandem mass spectra. It introduces approaches including spectral library search, protein database search, molecular fingerprinting, in silico spectral prediction, and de novo spectral annotation, covering applications in both proteomics and metabolomics. Key algorithms that underpin these approaches are discussed.
The second talk focuses on in silico spectral prediction and examines its potential and limitations for molecular identification. Mistle, a high-speed spectral search engine, is presented and evaluated for its ability to improve identification rates in large metaproteomics search spaces by searching complete in silico spectral libraries. Subsequently, FIORA, a novel graph neural network-based fragmentation algorithm, is introduced for in silico prediction of small-molecule metabolite spectra. Together, our findings demonstrate the potential of spectral prediction to aid computational identification of compounds and peptides.
Zeit & Ort
28.09.2026 | 14:30
Seminarraum 005
(Fachbereich Mathematik und Informatik, Takustr. 9, 14195 Berlin)
&
WebEx