| Timetable |
| Download Conference Program |
| Download All Abstracts |
| Zoom Access |
| Corporate Program |
Poster Presentations
Day 2, June 23（Mon.）
Room P (Maesato East, Foyer, Ocean Wing)
- 2P-AM-49
Nonnegative Tensor Factorization Enables Precursor-Peptide-Protein Deconvolution in Data-independent Acquisition Mass Spectrometry
(1Kyoto Univ. (Pharma), 2Kyoto Univ. (Engr), 3Kyoto Univ. (Info), 4NIBIOHN)
oJherico Geronca1, Kazuyoshi Yoshii2, Toshiyuki Tanaka3, Yasushi Ishihama1,4
Introduction
Bottom-up proteomics is a powerful approach for identifying and quantifying proteins by analyzing peptide fragments generated through enzymatic digestion. Data-independent acquisition (DIA) enables comprehensive proteome identification by fragmenting all precursor ions within a selected mass range. However, DIA generates complex signals that require robust data analysis strategies for accurate deconvolution. Conventional methods, like DIA-NN, rely on scoring to select the "best precursor," discarding a significant portion of the data and potentially reducing identification rates.
Methodology
We developed a novel deconvolution method based on Nonnegative Tensor Factorization, which resolves challenging chimeric signals in bottom-up proteomics by leveraging <I>a priori<I> product ion and peptide relative intensities.
Results
Our model successfully deconvolved product ions sharing the same <I>m/z<I> from two precursors, improving the Pearson correlation from 0.8622 to 1.0000, indicating perfect deconvolution. Additionally, when resolving shared peptide signals from two different proteins, the correlation improved from 0.9346 to 0.9936, demonstrating very high accuracy in separating peptide mixed signals.
Conclusion
Nonnegative Tensor Factorization is an effective method for deconvolving complex DIA signals, resolving overlapping product ions and shared peptide signals. By leveraging <I>a priori<I> relative intensities, it minimizes signal interference, improving identification accuracy and enhancing DIA-based proteomics.