Ggplot2

To facilitate model curation and analyzing pathways, Escher was used for visualizing the fluxes in the model's metabolic pathways. Escher enables the building of metabolic pathways using reactions, metabolites, and genes by contextualizing them in the organism's metabolism (King et al., 2015 (link)). The Escher Python package v1.7.1 (King et al., 2015 (link)) was also used to draw customized metabolic maps of C. canadensis 85B in Jupyter notebooks as it is compliant with COBRApy. Graphs for carbon source predictions were plotted with ggplot2 (Wickham, 2009 (link)) in R studio version 4.1.1 (RStudio Team, 2015 ).

For GO enrichment analysis on the biological processes domain, the hypergeometric test in the GOstats package (version 2.56.0) was applied. The analysis was restricted to gene sets containing 5–1,000 genes. Significant pathways were filtered by applying P value <0.05 and gene count/term >10. Further selection of relevant GO terms was based on sorting terms on their OddsRatios (the ratio of a GO term in the differently expressed genes list to the occurrence of this GO term in a universal gene list, obtained from org.Mm.eg.db [version 3.13.0]). In addition, GO term results were screened for enrichment of terms related to immune regulation. Selected top pathways were visualized using ggplot2 (version 3.3.5).
A web application for data searching and visualization was generated using the shiny package of Rstudio (https://shiny.rstudio.com), and the package ShinyCell for database creation (Ouyang et al., 2021 (link)).

Randomization tests were performed using the online tool PlotsOfDifferences (https://huygens.science.uva.nl/PlotsOfDifferences/) (Goedhart, 2019 (link)). Dot plots were generated using PlotsOfData (Postma and Goedhart, 2019 (link)). SuperPlots were generated using SuperPlotsofData (Lord et al., 2020 (link); Goedhart, 2021 (link)). Bar plots with visualized data points, time-series data, and density plots were generated using R (https://www.r-project.org/), Rstudio (Integrated Development for R. RStudio, Inc., Boston, MA. https://www.rstudio.com/) and ggplot2 (Wickham, 2016 ). The univariate Kolmogorov-Smirnov test was performed using Rstudio. Other statistical analyses were performed using Google sheets except for the one-sample t test which was performed using an online calculator (https://www.socscistatistics.com/tests/tsinglesample/default.aspx).

A minimum of three independent experiments (or animals) was used for all assays. Statistical analysis was carried out using Graphpad Prism software. Multiple data points were collected per mouse to generate the population mean and represent it as raincloud plots using ggplot2 in R-studio. These values were used to compare between control and mutant mice. Unpaired, two tailed Student’s t tests were used to calculate statistical significances, shown as p values, means and SD are shown. Plots were generated using Graphpad Prism and comparison was done at each fibre range.

RStudio software (RStudio Team, 2020 ) was used for data analysis. Normality and homoscedasticity of the data were assessed, and the data met these assumptions. The RStudio package “rsm” (Lenth, 2009 ) was used to analyse inflorescence yield and to generate three-dimensional and contour plots to represent the response surface. To improve the precision of yield estimates, the average yield of the five replicates in each treatment was used. Two sets of three surface and contour plots were created, each while holding one of the nutrient concentrations fixed at its centre point. These surface and contour plots, along with canonical analysis, were then used to determine the optimal rate of all three factors. Correlation analysis of yield and vegetative parameters was performed using the RStudio software package “ggplot2” (Wickham, 2016 ). To determine if there were differences in inflorescence cannabinoid content attributable to treatment, data from cannabinoid analysis was tested with a one-way ANOVA followed by Tukey’s HSD post hoc test.

The analysis of variance was performed for CSRFAB scores, slope and AUGCP predictors that take into account all the time harvests at once. R package lme4 [62 (link)] was used for statistical analysis. The model for the phenotypic data analysis was Y_ijk = μ + G_i + T_j + E_k + GT_ij + GE_ik + GTE_ijk + R_lk + e_ijkl; where μ is the total mean, Gi is the effect of the ith genotype, T_j is the effect of the j^th harvesting time, GT_ij is the effect of the interaction between the i^th genotype and the j^th harvesting time, GTE_ijk is the effect of the interaction between the i^th genotype and the j^th harvesting time in the k^th environment, E_k is the effect of the k^th environment, GE_ik is the interaction effect between the i^th genotype and the k^th environment, R_ik/ is the effect of the L^th block within the k^th environment, and e_ijk is a random error .
Correlation analyses between CSRFAB traits were performed using the cor function and Pearson method in R software.
Boxplots [63 ] were used to display a five-number data summary to better describe the variability within the data. The aesthetics with aes() function together with the geom_boxplot() layer both in ggplot2 (performed in Rstudio) provided a visualization of the data variability and dispersion.

PDX models were generated as described [31 (link)]. RNAseq was carried out on snap-frozen tissues from early passage (passage 1 to 3) PDX models, as described above. Expression of SULF1 and SULF2 was compared between patient and xenograft samples using a paired Wilcoxon rank sum test and visualized using the R package ggplot2 (v. 3.3.6) (RStudio, Auckland, New Zealand).