Over Eight Hundred Cannabis Strains Characterized by the Relationship
Total Page：16
File Type：pdf, Size：1020Kb
bioRxiv preprint doi: https://doi.org/10.1101/759696; this version postedClick September here 8,to 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rightsaccess/download;Manuscript;manuscrito_cannabis_final_JCR.do reserved. No reuse allowed without permission. Click here to view linked References 1 2 3 4 5 1 Over eight hundred cannabis strains characterized by the relationship between 6 7 2 their psychoactive effects, perceptual profiles, and chemical compositions 8 9 10 3 11 12 4 Alethia de la Fuente1,2,3, Federico Zamberlan1,3, Andrés Sánchez Ferrán4, Facundo Carrillo3,5, Enzo 13 14 5 Tagliazucchi1,3, Carla Pallavicini1,3,6* 15 16 17 6 18 19 7 1 Buenos Aires Physics Institute (IFIBA) and Physics Department, University of Buenos Aires, Buenos 20 21 8 Aires, Argentina. 22 23 9 2 Institute of Cognitive and Translational Neuroscience (INCYT), INECO Foundation, Favaloro University, 24 25 26 10 Buenos Aires, Argentina. 27 28 11 3 National Scientific and Technical Research Council (CONICET), Buenos Aires, Argentina. 29 30 12 4 Universidad Nacional de Tucumán, San Miguel de Tucumán, Argentina. 31 32 13 5 Applied Artificial Intelligence Lab, ICC, CONICET, Buenos Aires, Argentina. 33 34 6 35 14 Grupo de Investigación en Neurociencias Aplicadas a las Alteraciones de la Conducta, FLENI-CONICET, 36 37 15 Buenos Aires, Argentina. 38 39 40 16 41 42 17 *Corresponding author: [email protected] 43 44 18 45 46 19 47 48 49 20 50 51 21 52 53 22 54 55 23 56 57 58 24 59 60 25 61 62 63 64 65 1 bioRxiv preprint doi: https://doi.org/10.1101/759696; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 2 3 4 1 Abstract 5 6 7 2 Background: Commercially available cannabis strains have multiplied in recent years as a consequence of 8 9 3 regional changes in legislation for medicinal and recreational use. Lack of a standardized system to label 10 11 4 plants and seeds hinders the consistent identification of particular strains with their elicited psychoactive 12 13 5 effects. The objective of this work was to leverage information extracted from large databases to improve 14 15 6 the identification and characterization of cannabis strains. 16 17 18 7 Methods: We analyzed a large publicly available dataset where users freely reported their experiences with 19 20 8 cannabis strains, including different subjective effects and flavour associations. This analysis was 21 22 9 complemented with information on the chemical composition of a subset of the strains. Both supervised 23 24 10 and unsupervised machine learning algorithms were applied to classify strains based on self-reported and 25 26 27 11 objective features. 28 29 12 Results: Metrics of strain similarity based on self-reported effect and flavour tags allowed machine learning 30 31 13 classification into three major clusters corresponding to Cannabis sativa, Cannabis indica, and hybrids. 32 33 14 Synergy between terpene and cannabinoid content was suggested by significative correlations between 34 35 36 15 psychoactive effect and flavour tags. The use of predefined tags was validated by applying semantic 37 38 16 analysis tools to unstructured written reviews, also providing breed-specific topics consistent with their 39 40 17 purported medicinal and subjective effects. While cannabinoid content was variable even within individual 41 42 18 strains, terpene profiles matched the perceptual characterizations made by the users and could be used to 43 44 45 19 predict associations between different psychoactive effects. 46 47 20 Conclusions: Our work represents the first data-driven synthesis of self-reported and chemical information 48 49 21 in a large number of cannabis strains. Since terpene content is robustly inherited and less influenced by 50 51 22 environmental factors, flavour perception could represent a reliable marker to predict the psychoactive 52 53 23 effects of cannabis. Our novel methodology contributes to meet the demands for reliable strain classification 54 55 56 24 and characterization in the context of an ever-growing market for medicinal and recreational cannabis. 57 58 25 59 60 26 Keywords: Cannabis strains, terpenes, cannabinoids, flavour, chemotypes, subjective reports. 61 62 63 64 65 2 bioRxiv preprint doi: https://doi.org/10.1101/759696; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 2 3 4 1 Background 5 6 7 2 Cannabis indica and Cannabis sativa have been used in traditional medicine for millennia around the world, 8 9 3 as well as a source for fiber and food (Mechoulam, 2019; Russo, 2011; Russo et al., 2008). In the past 10 11 4 century, Western medicine has gone a long way to find specific medications for many afflictions 12 13 5 traditionally treated with cannabis-derived products, and the recreational use of marihuana for its 14 15 6 psychoactive properties emerged as the main motivation for its cultivation and consumption (Clarke & 16 17 18 7 Merlin, 2013). This and other factors led to the inclusion of cannabis as a Schedule 1 controlled substance, 19 20 8 a category reserved for compounds with “no currently accepted medical use" according to the Food and 21 22 9 Drug Administration (U.S. Food and Drug Administration, 2015), despite its long history in the treatment 23 24 10 of diverse illnesses, symptoms and conditions (Clarke & Merlin, 2013). 25 26 27 11 Recently, regional changes in legislation made marihuana and other cannabis-derived products available 28 29 12 for medicinal and recreational use (Bifulco & Pisanti, 2015; Fischer, Ala-Leppilampi, Single, & Robins, 30 31 13 2010; Graham, 2015; Hetzer & Walsh, 2014). It is expected that through resilient patient/consumer activism 32 33 14 and increasing scientific evidence supporting the medicinal use of cannabis, this phenomenon will continue 34 35 36 15 to rise gradually in more countries (Hazekamp, Tejkalová, & Papadimitriou, 2016). Market growth for 37 38 16 marihuana has been dramatic in some countries; for instance, in the United States sales reached $6.7 billion 39 40 17 in 2016, with 30% growth year-over-year, representing the second largest cash crop, with total worth over 41 42 18 $40 billion (Adams, 2019; Robinson, 2017). These sudden changes created novel problems for users, as 43 44 45 19 cannabis cultivators transition towards legal business models, yet without a world-wide standard for their 46 47 20 products. Cannabis dispensaries offer dry cannabis flowers or buds (Gilbert & DiVerdi, 2018), extracts and 48 49 21 essential oils (Permanente & Care, 2008) and various edibles (Weedmaps, n.d.); however, since in most 50 51 22 countries these products remain illegal, there are no international agreements to regulate their quality or 52 53 23 chemical content. 54 55 56 24 The development of standards is further complicated by the heterogeneous chemical composition inherent 57 58 25 to cannabis. Plants contain over 400 compounds, including more than 60 cannabinoids, the main active 59 60 26 molecules being tetrahydrocannabinol (THC) and cannabidiol (CBD). These two cannabinoids were often 61 62 63 64 65 3 bioRxiv preprint doi: https://doi.org/10.1101/759696; this version posted September 8, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 2 3 4 1 considered the only chemicals involved in the medicinal properties and psychoactive effects associated with 5 6 7 2 cannabis, and remain the only ones screened when evaluating strain chemotypes (De Meijer, Hammond, & 8 9 3 Sutton, 2009; Fetterman et al., 1971; Hazekamp et al., 2016; Nie, Henion, & Ryona, 2019; UNODC, 1968). 10 11 4 However, increasing evidence supports the relevance of terpenes and terpenoids, molecules responsible for 12 13 5 the flavour and scent of the plants, both as synergetic to cannabinoids and as active compounds by 14 15 6 themselves (Henry, 2017; Hillig, 2004; Nuutinen, 2018; Russo, 2011). Flavours have predictive value at 16 17 18 7 strain level (Gilbert & DiVerdi, 2018) that may not be superseded by the determination of the species, or 19 20 8 even by quantification of THC and CBD content (Jikomes & Zoorob, 2018). Terpenes are widely used as 21 22 9 biochemical markers in chemosystematics studies to characterize plant samples due to the fact that they are 23 24 10 strongly inherited and relatively unaffected by environmental factors (Aizpurua-Olaizola et al., 2016; 25 26 27 11 Casano, Grassi, Martini, & Michelozzi, 2011; Hillig, 2004). Cannabinoid content, on the other hand, can 28 29 12 vary greatly among generations of the same strain, and also due to the sex, age and part of the plant 30 31 13 (Fetterman et al., 1971; Hazekamp et al., 2016). 32 33 14 In this work, we combined different sources of data for the classification of cannabis strains, linking both 34 35 36 15 self-reports of psychoactive effects and flavour profiles with information obtained from experimental 37 38 16 assays of cannabinoid and terpene content. Our analysis comprised 887 different strains and was based on 39 40 17 a large sample (>100.000) of user reviews publicly available at the website Leafly (www.leafly.com). The 41 42 18 reports contained unstructured written reviews of experiences