Determination of agronomically important genes in cannabis plant by bioinformatics methods

This study morphologically investigated the effects of brassinosteroids (BR) and BR inhibitor brassinazole (BRZ) in the local cannabis (Cannabis sativa L.) cultivars "Narlı" and "Vezir". On the other hand, a database of Cannabis sativa-specific TF proteins was created, and a total of 1,323 TF genes belonging to 57 TF families identified with peptide domains conserved in the Cannbio-2 genome assembly were identified and characterized using bioinformatics methods. Chromosomal locations, gene structures, physicochemical properties of their peptides, conserved peptide motifs, phylogenetic relationships, predicted 3D peptide structures, sequence information were determined by bioinformatics analysis and uploaded to the database. A hybrid deep learning model was developed to predict all TFs encoded in the Cannabis sativa genome. This model combines Word2Vec based sequence representation with Convolutional Neural Networks (CNN), Bidirectional Gated Recurrent Unit (BiGRU) and Attention mechanism. The model was trained on a dataset of 13,499 peptides and achieved 97.80% accuracy and 97.74% F1-score, outperforming existing methods in classifying these proteins. Narlı and Vezir varieties were grown under controlled conditions. BR and BRZ applied at different developmental stages significantly affected plant height and technical stem length parameters. Male plants responded particularly stronger to BR treatment compared to females. Male plants responded more strongly to BR treatments than females, indicating an increase in potential fiber yield. BR treatments increased growth parameters such as plant height and technical stem length, while BRZ treatment suppressed these effects. These findings offer new horizons for hemp agriculture and potential uses of brassinosteroids in agricultural applications. Furthermore, the bioinformatics tools and deep learning-based classification model developed in this study enabled the comprehensive and highly accurate identification of TF genes in the cannabis genome. This approach provides a strategic starting point for future studies that will elucidate the genetic basis of cannabis in fiber development, stress tolerance and metabolic regulation.