Objective:

To characterize flavor-related chemicals in five edible bolete species and assess whether spectroscopy and machine learning could support rapid screening.

Approach:

• Samples: Researchers analyzed 82 fruiting bodies collected in southwest China, combined into 30 samples with six biological replicates per species.
• Chemical analysis: Liquid chromatography–tandem mass spectrometry measured amino acids and organic acids; headspace solid-phase microextraction gas chromatography–mass spectrometry measured volatile compounds.
• Additional analyses: The team compared compound levels with published taste and odor thresholds, modeled possible odor-receptor binding, and tested infrared spectroscopy with machine learning.

Key Findings:

• The analysis identified 68 amino acids and related metabolites, 57 organic acids, and 773 volatile compounds; chemical profiles separated the five species.
• Glycine, arginine, aspartic acid, lysine, and fumaric acid were among the compounds distinguishing taste profiles. Sweet and umami amino acids contributed most.
• Six volatile compounds were proposed as aroma markers, spanning pyrazines, an ester, an aldehyde, and a furanone associated with varied odor descriptions.
• Pathway analysis linked chemical differences to the tricarboxylic acid and urea cycles, but these relationships were not experimentally confirmed.
• Computer modeling predicted that several aroma compounds could bind to OR1A1; this did not establish how people perceive the aromas.
• The strongest infrared spectroscopy and machine-learning model predicted selected flavor-compound concentrations with coefficients of determination of at least 0.95.

Interpretation:

The authors report that chemical profiling distinguished the species and that infrared spectroscopy with machine learning showed potential for faster initial screening, while mass spectrometry provided detailed chemical measurements.

Limitations:

• The study included a limited number of samples and collection sites.
• Habitat differences may have influenced the chemical profiles.
• Metabolic pathway relationships and predicted receptor binding were not experimentally confirmed.

Sources:

• Article context provided by the user