In vsritter/BiosCoreHub: NC TraCS Bios Core sharing hub

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(BiosCoreHub)

Use RStudio!

• Latest versions of RStudio have many features that can help you writing your reports

  • Markdown editor (new!)
  • Templates for reports, papers, and presentations
  • Spell checking
  • Automation and reproducibility

Project directory structure (optional)

• Each request gets a new Rstudio project folder
• Project's folder has subdirectories for general documents, data, scripts, and report (the report folder is created using the BiosCoreHub package).

Workflow (optional)

• Complicated analysis with multiple scripts are saved in the R directory and sourced when writing the RMarkdown script

  • Pros: faster .Rmd compilation (can use knitr option cache = TRUE)
  • Pros: cleaner .Rmd file
  • Cons: .Rmd is not a standalone report anymore
  • Cons: can be complicated to manage which codes to use on your report in case of multiple iterations

• Example:

knitr::opts_chunk$set(echo = FALSE, fig.align = "center")

# Load the packages you need
library(tidyverse)
library(flextable)

source('./R/analysis.R')

Table 1 and Results

• Package gtsummary has great capabilities
• Remember to label your variables for nice outputs (e.g. labelled package)

{width="800"}

Examples (straight from gtsummary's website here)

Summary Table

Use tbl_summary() to summarize a data frame.

library(gtsummary)
# make dataset with a few variables to summarize
trial2 <- trial %>% select(age, grade, response, trt)

# summarize the data with our package
table1 <- tbl_summary(trial2)

table1

There are many customization options to add information (like comparing groups) and format results (like bold labels) in your table. See the tbl_summary() tutorial for many more options, or below for one example.

table2 <- 
  tbl_summary(
    trial2,
    by = trt, # split table by group
    missing = "no" # don't list missing data separately
  ) %>%
  add_n() %>% # add column with total number of non-missing observations
  add_p() %>% # test for a difference between groups
  modify_header(label = "**Variable**") %>% # update the column header
  bold_labels() 

table2

Regression Models

Use tbl_regression() to easily and beautifully display regression model results in a table. See the tutorial for customization options.

mod1 <- glm(response ~ trt + age + grade, trial, family = binomial)

t1 <- tbl_regression(mod1, exponentiate = TRUE)

t1

Side-by-side Regression Models

You can also present side-by-side regression model results using tbl_merge()

library(survival)

# build survival model table
t2 <-
  coxph(Surv(ttdeath, death) ~ trt + grade + age, trial) %>%
  tbl_regression(exponentiate = TRUE)

# merge tables 
tbl_merge_ex1 <-
  tbl_merge(
    tbls = list(t1, t2),
    tab_spanner = c("**Tumor Response**", "**Time to Death**")
  )

tbl_merge_ex1

Save Individual Tables

gtsummary tables can also be saved directly to file as an image, RTF, LaTeX, and Word file.

tbl %>%
  as_flex_table() %>%
  flextable::save_as_docx()

Small fonts and themes

# If you want small fonts
table1 %>%
  as_gt() %>%
  gt::tab_options(table.font.size = "small")

# JAMA theme
theme_gtsummary_journal(journal = "jama")

vsritter/BiosCoreHub documentation built on Dec. 23, 2021, 4:13 p.m.