The local-balanced model for improved machine learning outcomes on mass spectrometry data sets and other instrumental data

Abstract

One unifying challenge when classifying biological samples with mass spectrometry data is overcoming the obstacle of sample-to-sample variability so that differences between groups, such as between a healthy set and a disease set, can be identified. Similarly, when the same sample is re-analyzed under identical conditions, instrument signals can fluctuate by more than 10%. This signal inconsistency imposes difficulties in identifying subtle differences across a set of samples, and it weakens the mass spectrometrist’s ability to effectively leverage data in domains as diverse as proteomics, metabolomics, glycomics, and imaging. We selected challenging data sets in the fields of glycomics, mass spectrometry imaging, and bacterial typing to study the problem of within-group signal variability and adapted a 30 year old statistical approach to address the problem. The solution, “local-balanced model,” relies on using balanced subsets of training data to classify test samples. This analysis strategy was assessed on ESI-MS data of IgG-based glycopeptides and MALDI-MS imaging data of endogenous lipids, and MALDI-MS data of bacterial proteins. Two preliminary examples on non-mass spectrometry data sets are also included to show the potential generality of the method outside the field of MS analysis. We demonstrate that this approach is superior to simple normalization methods, generalizable to multiple mass spectrometry domains, and potentially appropriate in fields as diverse as physics and satellite imaging. In some cases, improvements in classification can be dramatic, with accuracy escalating from 60% with normalization alone to over 90% with the additional development described herein.

An analytical study on the identification of N-linked glycosylation sites using machine learning model

“…It is observed from the selected articles that most of the authors ( Akmal, Rasool & Khan, 2017 ; Taherzadeh et al, 2019 ; Liu et al, 2019 ; Lundstrøm et al, 2022 ; Burkholz, Quackenbush & Bojar, 2021 ; Kotidis & Kontoravdi, 2020 ; Antonakoudis et al, 2021 ; Hwang et al, 2020 ; Dimeglio et al, 2020 ; Dobson, Zeke & Tusnády, 2021 ; Ilyas et al, 2019 ; Chen et al, 2021 ) used the Artificial Neural Network (ANN) or the variant of ANN such as Deep ANN, Graph NN, Convolution NN and Recurrent NN. The second most used algorithm is Support Vector Machine (SVM) used by authors ( Tran, Pham & Ou, 2021 ; Pitti et al, 2019 ; Wang et al, 2017 ; Desaire, Patabandige & Hua, 2021 ; Qiu et al, 2018 ) and remaining authors used Random Forest, XGBOOST, Baysen Network, Regression Classifier, Radial Base Function and some used customized method as mention in Table 11 .…”

“…It has also been analysed the research article ( Taherzadeh et al, 2019 ; Le, Sandag & Ou, 2018 ; Ruiz-Blanco et al, 2017 ) in which web server has provided and present the accuracy above 90% used the Jrip Classifier, DNN, SVM and RBF algorithm. The authors ( Akmal, Rasool & Khan, 2017 ; Tran, Pham & Ou, 2021 ; Hwang et al, 2020 ; Magaret et al, 2019 ; Desaire, Patabandige & Hua, 2021 ) who proposed prediction model without providing the webserver and also have accuracy above 90% used ANN, SVM, DNN and RF algorithms. The researchers can use these algorithms to improve the performance of N-linked prediction model or any PTM site identification model.…”

Leveraging R (LevR) for fast processing of mass spectrometry data and machine learning: Applications analyzing fingerprints and glycopeptides

“…Here, we present the pipeline and show its utility using two different data sets. The output is compatible with machine learning strategies, like the Aristotle Classifier (Hua et al, 2019;Desaire and Hua, 2020;Hua et al, 2020;Desaire et al, 2021), which makes use of the many features within a spectrum that can all contribute to identifying a disease state. This tool will aid mass spectrometrists who have previously lacked accessibility to apply machine learning strategies to their data sets.…”