exohnsc

Code for the analyses and figures in:

Dhakal A, Upadhyay R, Wheeler C, Hoyd R, Karivedu V, Gamez ME, Valentin S, Vanputten M, Bhateja P, Bonomi M, Konieczkowski DJ, Baliga S, Mitchell DL, Grecula JC, Blakaj DM, Denko NC, Jhawar SR, Spakowicz D. Association between Tumor Microbiome and Hypoxia across Anatomic Subsites of Head and Neck Cancers. International Journal of Molecular Sciences. 2022;23(24):15531. doi: 10.3390/ijms232415531. PMID: 36555172.

Regenerate the manuscript figures

Figure 1 and Figure 2A–C are drawn by the R Markdown files in scripts/. From the repository root, create the output directory and render all four. rmarkdown::render() knits each file with scripts/ as the working directory, which is what the ../data and ../figures paths expect.

install.packages(c(
  "rmarkdown", "tidyverse", "colorspace", "scales", "svglite",
  "ggdist", "RColorBrewer", "ggforce", "patchwork", "ggrepel"
))

mkdir -p figures
Rscript -e 'for (f in c(
  "scripts/fig1_buffa.Rmd",
  "scripts/fig2A_stackedBar.Rmd",
  "scripts/fig2B_volcano.Rmd",
  "scripts/fig2C_sigMicrobes.Rmd"
)) rmarkdown::render(f)'

Knitting the same files from RStudio with exohnsc.Rproj open is equivalent.

Script	Manuscript figure	Output
scripts/fig1_buffa.Rmd	Figure 1. Buffa hypoxia scores across TCGA cancer types, with HNSC split by subsite	figures/fig1_buffa-cancer-raincloud.png, figures/fig1_buffa-cancer-raincloud.svg
scripts/fig2A_stackedBar.Rmd	Figure 2A. Phylum-level relative abundance by tumor, faceted by subsite and ordered by Proteobacteria	figures/fig1_stackedbar.png, figures/fig1_stackedbar.svg
scripts/fig2B_volcano.Rmd	Figure 2B. Effect size versus unadjusted p-value for each subsite. Taxa with p < 0.01 and an absolute effect size above 0.5 are labeled	figures/fig1_volcano_faceted.png
scripts/fig2C_sigMicrobes.Rmd	Figure 2C. Effect sizes for the species named in the script, including Pseudomonas, Actinomyces, Sulfurimonas, Filifactor alocis, Porphyromonas gingivalis, and Helicobacter pylori	figures/fig1_effectsize_selectmics.png

The ggsave() calls for Figure 2 still use a fig1_ filename prefix.

Data each script reads

Figure 1, Figure 2B, and Figure 2C use files that are already in data/:

• data/buffa-mitophagy_tcga-cancer.RData and data/location.csv for Figure 1. The R data file holds Buffa scores for TCGA cancer types. location.csv labels HNSC samples as oral cavity, oropharynx, or larynx/hypopharynx.
• data/modelling_binom_OralCavity.csv, data/modelling_binom_Oropharynx.csv, and data/modelling_binom_LarynxHypopharinx.csv for Figure 2B and Figure 2C. Each file is a table of binomial-model coefficients (term, estimate, std.error, statistic, p.value) for microbe relative abundance against a high-versus-low Buffa score.

Figure 2A also reads data/hnsc_buffa.csv and two cohort files that are not in this repository. The paths are hard-coded in scripts/fig2A_stackedBar.Rmd:

/fs/ess/PAS1695/projects/HNSC/data/drake-output/7-15-2021/7-15-2021_tcga_exora-with-taxonomy.csv
/fs/ess/PAS1695/projects/HNSC/data/drake-output/7-15-2021/7-15-2021_tcga_clinical.csv

Point those two read.csv() calls at local copies before rendering Figure 2A. The taxonomy table supplies exogenous-sequence relative abundances. The clinical table supplies primary_site, which the script maps to oral cavity, oropharynx, and larynx/hypopharynx.