MALDI(+) FT-ICR Mass Spectrometry (MS) Combined with Machine Learning toward Saliva-Based Diagnostic Screening for...

Area of Science:

• Analytical Chemistry
• Biotechnology
• Computational Biology

Background:

• Rapid identification of respiratory viruses is crucial for pandemic control.
• Machine learning applied to mass spectrometry data shows potential for classifying SARS-CoV-2 positive samples.

Purpose of the Study:

• To develop a fast and effective saliva-based screening method for COVID-19 using MALDI FT-ICR MS and support vector machine (SVM).

Main Methods:

• Optimization of saliva sample preparation involving trypsin digestion.
• Utilizing matrix-assisted laser desorption/ionization Fourier-transform ion cyclotron resonance mass spectrometry (MALDI FT-ICR MS) for data acquisition.
• Developing and evaluating SVM models (SVM1, SVM2) with RT-PCR confirmed samples (97 positive, 52 negative).

Main Results:

• Optimal sample preparation involved 2h trypsin digestion and MALDI analysis at 1 M.
• SVM models demonstrated high performance: 100% accuracy in calibration and 95.6% (SVM1) / 86.7% (SVM2) in testing.
• SVM1 achieved 0% false negative rate (FNR) with 780 variables; SVM2 had 3% FNR with two variables.

Conclusions:

• The proposed MALDI FT-ICR MS and SVM methodology is a promising tool for aiding COVID-19 screening.
• The method offers a rapid and effective approach for identifying SARS-CoV-2 in patient saliva samples.