Algorithmic Learning for Auto-deconvolution of GC-MS Data to Enable Molecular Networking within GNPS

• Published in: bioRxiv
• Main Authors: Aksenov, Alexander, Laponogov, Ivan, Zhang, Zheng, Doran, Sophie, Belluomo, Ilaria, Veselkov, Dennis, Bittremieux, Wout, Louis Felix Nothias, Nothias-Esposito, Melissa, Maloney, Katherine N, Misra, Biswapriya Biswavas, Melnik, Alexey V, Jones, Kenneth L, Dorrestein, Kathleen, Morgan Panitchpakdi, Ernst, Madeleine, Justin Jj Van Der Hooft, Gonzalez, Mabel, Carazzone, Chiara, Amezquita, Adolfo, Callewaert, Chris, Morton, Jamie, Quinn, Robert Andrew, Bouslimani, Amina, Andrea Albarracin Orio, Petras, Daniel, Smania, Andrea Maria, Couvillion, Sneha P, Burnet, Meagan C, Nicora, Carrie D, Zink, Erika, Metz, Thomas O, Artaev, Viatcheslav, Humston-Fulmer, Elizabeth, Gregor, Rachel, Meijler, Michael M, Mizrahi, Itzhak, Stav Eyal, Anderson, Brooke, Dutton, Rachel J, Lugan, Raphael, Pauline Le Boulch, Guitton, Yann, Prevost, Stephanie, Poirier, Audrey, Gaud Dervilly, Bruno Le Bizec, Fait, Aharon, Sikron, Noga, Song, Chao, Gashu, Kelem, Coras, Roxana, Guma, Monica, Manasson, Julia, Scher, Jose U, Barupal, Dinesh K, Saleh Alseekh, Fernie, Alisdair, Fernie, Alisdair R, Mirnezami, Reza, Vasiliou, Vasilis, Schmid, Robin, Borisov, Roman S, Kulikova, Larisa N, Knight, Rob, Wang, Mingxun, Hanna, George, Dorrestein, Pieter C, Veselkov, Kirill
• Format: Paper
• Language: English
• Published: Cold Spring Harbor Cold Spring Harbor Laboratory Press 14.01.2020; Cold Spring Harbor Laboratory
• Edition: 1.1
• Subjects: Bioinformatics; Fourier analysis; Gas chromatography; Gastric cancer; Ions; Learning algorithms; Machine learning; Mass spectrometry; Mass spectroscopy; Natural products; Scientific imaging; Social organization
• ISSN: 2692-8205; 2692-8205
• DOI: 10.1101/2020.01.13.905091

Gas chromatography-mass spectrometry (GC-MS) represents an analytical technique with significant practical societal impact. Spectral deconvolution is an essential step for interpreting GC-MS data. No public GC-MS repositories that also enable repository-scale analysis exist, in part because deconvolution requires significant user input. We therefore engineered a scalable machine learning workflow for the Global Natural Product Social Molecular Networking (GNPS) analysis platform to enable the mass spectrometry community to store, process, share, annotate, compare, and perform molecular networking of GC-MS data. The workflow performs auto-deconvolution of compound fragmentation patterns via unsupervised non-negative matrix factorization, using a Fast Fourier Transform-based strategy to overcome scalability limitations. We introduce a "balance score" that quantifies the reproducibility of fragmentation patterns across all samples. We demonstrate the utility of the platform with breathomics analysis applied to the early detection of oesophago-gastric cancer, and by creating the first molecular spatial map of the human volatilome. Footnotes * https://www.youtube.com/watch?v=yrru-5nrsdk&feature=youtu.be * https://www.youtube.com/watch?v=MblruOSglgI&feature=youtu.be * https://www.youtube.com/watch?v=iX03r_mGi2Q&feature=youtu.be * https://www.youtube.com/watch?v=mv-fw2zSgss&feature=youtu.be * https://www.youtube.com/watch?v=nUhCZ9LwoM4&feature=youtu.be * https://www.youtube.com/watch?v=_PehOiBqzzY&feature=youtu.be