Please note: This seminar will take place in DC 1304.
Haixu Tang, Professor, Computer Science
Director, Data Science Academic Programs
Luddy School of Informatics, Computing, and Engineering
Tandem mass spectrometry (MS/MS) plays a crucial role in precision health by providing detailed structural information for peptides and small molecules, enabling the analysis of complex biological samples and driving personalized medicine initiatives. Computational methods, particularly those employing deep learning, become increasingly important for extracting meaningful insights from these rich datasets.
This talk explores the application of deep learning algorithms to decipher complex patterns within MS/MS spectra for improved molecular identification and its subsequent impact on personalized medicine. We present PepNet, a fully convolutional neural network designed for de novo peptide sequencing. PepNet facilitates de novo peptide sequencing from immunopeptidomic datasets, generating a library of potential neoantigenic peptides for personalized cancer immunotherapy. Turning to small molecule analysis, we introduce a dual-stage deep learning framework. First, we introduce Mol3DMS, a point-based deep neural network predicting small molecule MS/MS spectra from 3D conformations. We demonstrate its utility in untargeted metabolomics, where predicted spectra are used to search experimental MS/MS data for metabolite identification. We present FIDDLE (Fast Identification of chemical Formulas from mass spectra using Deep LEarning), which enables the rapid and accurate determination of molecular formulas directly from raw spectra.
Collectively, these AI-driven methodologies empower predictive models to discover metabolite biomarkers, which can be further leveraged to develop predictive deep learning models for disease prognosis, treatment effectiveness, and ultimately, more precise and personalized healthcare interventions.
Bio: Haixu Tang is a Professor in Computer Science and the Director of Data Science Academic Programs in the Luddy School of Informatics, Computing, and Engineering. He is interested in algorithmic and machine learning problems in computational biology, in particular in genomics, proteomics and metabolomics. He is the recipient of the NSF Early Career Award in 2007, and the Outstanding Junior Faculty Award from Indiana University in 2009.