Deriving three one dimensional NMR spectra from a single experiment through machine learning.

Area of Science:

• Metabolomics
• Analytical Chemistry
• Biochemistry

Background:

• Nuclear Magnetic Resonance (NMR) spectroscopy is vital for analyzing complex biological mixtures.
• NMR offers detailed molecular insights and preserves sample integrity, crucial for metabolomics.
• Standard NMR techniques like NOESY, CPMG, diffusion-edited, and JRES provide complementary data but require significant time for high-throughput studies.

Purpose of the Study:

• To develop a machine learning model for predicting NMR spectra.
• To streamline NMR-based metabolomics analysis by reducing experimental time and resource requirements.
• To demonstrate the feasibility of predicting CPMG, diffusion-edited, and JRES spectra from NOESY spectra using serum samples.

Main Methods:

• Utilized a machine learning approach to predict NMR spectra.
• Leveraged Nuclear Overhauser Effect SpectroscopY (NOESY) spectra as input data.
• Applied the method to serum samples for metabolomic analysis.

Main Results:

• Successfully predicted CPMG, diffusion-edited, and JRES spectra from NOESY spectra.
• Demonstrated a streamlined and efficient method for NMR-based metabolomics.
• Validated the approach using serum samples.

Conclusions:

• The proposed machine learning strategy significantly enhances efficiency in NMR-based metabolomics.
• This method reduces the need for acquiring multiple NMR spectra, saving time and resources.
• The approach holds promise for accelerating high-throughput metabolomic analyses.