Classification of cannabis strains in the Canadian market with discriminant analysis of principal components using genome-wide single nucleotide polymorphisms

Abstract

The cannabis community typically uses the terms “Sativa” and “Indica” to characterize drug strains with high tetrahydrocannabinol (THC) levels. Due to large scale, extensive, and unrecorded hybridization in the past 40 years, this vernacular naming convention has become unreliable and inadequate for identifying or selecting strains for clinical research and medicinal production. Additionally, cannabidiol (CBD) dominant strains and balanced strains (or intermediate strains, which have intermediate levels of THC and CBD), are not included in the current classification studies despite the increasing research interest in the therapeutic potential of CBD. This paper is the first in a series of studies proposing that a new classification system be established based on genome-wide variation and supplemented by data on secondary metabolites and morphological characteristics. This study performed a whole-genome sequencing of 23 cannabis strains marketed in Canada, aligned sequences to a reference genome, and, after filtering for minor allele frequency of 10%, identified 137,858 single nucleotide polymorphisms (SNPs). Discriminant analysis of principal components (DAPC) was applied to these SNPs and further identified 344 structural SNPs, which classified individual strains into five chemotype-aligned groups: one CBD dominant, one balanced, and three THC dominant clusters. These structural SNPs were all multiallelic and were predominantly tri-allelic (339/344). The largest portion of these SNPs (37%) occurred on the same chromosome containing genes for CBD acid synthases (CBDAS) and THC acid synthases (THCAS). The remainder (63%) were located on the other nine chromosomes. These results showed that the genetic differences between modern cannabis strains were at a whole-genome level and not limited to THC or CBD production. These SNPs contained enough genetic variation for classifying individual strains into corresponding chemotypes. In an effort to elucidate the confused genetic backgrounds of commercially available cannabis strains, this classification attempt investigated the utility of DAPC for classifying modern cannabis strains and for identifying structural SNPs.

On offer to Ontario consumers three years after legalization: A profile of cannabis products, cannabinoid content, plant type, and prices

“…There were many offers of blend or hybrid options available in every product category. The taxonomical classification of cannabis into indica and sativa are mostly based on marketing considerations as they do not exist in nature due to historical cross-breeding and misuse of nomenclature ( 50 , 51 ). However, cannabis users report differential subjective experiences between indica- and sativa-dominant products with greater preference for using indica in the evening while reporting feeling “relaxed, sleepy/tired” and sativa during the day while reporting feeling “alert/energized” ( 52 ).…”

Divergent effects of oral cannabis oil extracts marketed as C. indica or C. sativa on exertion of cognitive effort in rats.

“…However, the same study also concluded that neither the name associated with a sample nor its reported ancestry was a reliable predictor of its genetic identity (Sawler et al, 2015). Two very recent large-scale SNP analyses confirmed that the vernacular label of Sativa and Indica could only poorly account for the overall genetic relatedness of different cultivars (Jin et al, 2021; Watts et al, 2021). While this is frustrating from a scientific perspective, recent data highlight the importance of using cannabis formulations in scientific experiments which match those on the market; although again little genetic justification for the labeling of cultivars as Indica, Sativa, or hybrid was observed, the “research-grade cannabis” supplied by the National Institute on Drug Abuse was genetically distinct from those available from retail dispensaries (Schwabe et al, 2021), making its utility in scientific investigations questionable.…”

Examining population structure across multiple collections of Cannabis

“…Previous work has used various reference genomes [18] [24] [31] [68] and this reflects the current predicament within the industry where standards are still in development. The cs10 reference genome is the most complete assembly to date, but it contains non-functional THCAS gene sequences.…”