Pipeline for High-Throughput Modeling of Marijuana and Hemp Extracts

Area of Science:

• Analytical Chemistry
• Chemometrics
• Pharmacognosy

Background:

• Increasing consumption and regulation of cannabis products necessitate robust authentication methods.
• Distinguishing between marijuana and hemp is crucial for quality control and regulatory compliance.
• Existing methods may lack the efficiency for high-throughput screening of cannabis botanical extracts.

Purpose of the Study:

• To develop and evaluate a two-stage pipeline for high-throughput screening and chemotyping of cannabis spectra.
• To accurately authenticate and differentiate between marijuana and hemp samples.
• To classify cannabis spectra into distinct chemotypes for recognizing pharmacological properties and cultivars.

Main Methods:

• A two-stage pipeline was implemented, starting with an automatic soft independent modeling of class analogy (aSIMCA) for initial classification (marijuana vs. hemp) and novel spectra rejection.
• The second stage employed multivariate classifiers—fuzzy rule building expert system (FuRES), super partial least-squares-discriminant analysis (sPLS-DA), and support vector machine tree type entropy (SVMtreeH)—for chemotyping.
• Spectra data from proton nuclear magnetic resonance, mass, and ultraviolet analyses were utilized for evaluation.

Main Results:

• The aSIMCA model demonstrated high efficiency and efficacy in classifying spectra as either marijuana or hemp.
• The second-stage classifiers successfully performed chemotyping, identifying distinct chemotypes within the cannabis samples.
• Overall pipeline accuracy exceeded 95% across various spectral data sets, confirming its discriminant ability.

Conclusions:

• The developed two-stage pipeline offers a promising solution for automated, high-throughput screening and chemotyping of cannabis botanical products.
• The method provides accurate authentication and classification, supporting quality assurance in the manufacturing of cannabis products.
• This approach has potential applications beyond cannabis, extending to the analysis of other botanical products.