Symposium Sessions
Day 1, June 10（Wed.） 15:52-16:14 Room C (4F 413)
- 1C-S2-1552
Development of Tools for Visualization and Deconvolution of product ion spectra
(HMT)
oHiroyuki Yamamoto
Mass spectrometry (MS) is an indispensable technology in omics research, and product ion spectra obtained by tandem MS (MS/MS) play an important role in compound identification and annotation. In this presentation, we introduce our recent studies, spectral visualization based on analysis workflows developed in single-cell genomics, deconvolution methods for separating mixed spectra, and the application of artificial intelligence. For spectral visualization, product ion spectra are converted into a binned data matrix, followed by dimensionality reduction using principal component analysis (PCA), and visualized using UMAP. These methods are implemented in the R package MSplusR. For large-scale spectral datasets, we also developed the Python library LargeScaleMSPy, which performs incremental PCA with chunked HDF5 processing and UMAP. In addition, the R package MS2DecR implements deconvolution methods for spectra derived from data independent acquisition, including approaches based on multivariate curve resolution–alternating least squares. We further discuss future prospects for AI-driven automated data processing.