In oucru-biostats/C306: Collection of useful R codes for data analysis from the Biostatistics group at OUCRU

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

C306

This package provides a collection of useful R codes for data analysis from the Biostatistics group at OUCRU-HCM.

Installation

You can install C306 from github with:

# install.packages("devtools")
devtools::install_github("oucru-biostats/C306@feature-rbind")

List of functions

Main functions

These functions' purpose are to generate report tables, outputting a flextable or a intermediate object of class ss_table for post-hoc manipulation.

• sstable.baseline function to generate baseline table 1.
• sstable.survcomp function to generate dual arm descriptive and inferential analysis for time-to-event outcome, supporting Cox pH and Restricted mean surival time models
• sstable.survcomp.subgroup extending sstable.survcomp for subgroup analyses
• sstable.ae function to generate adverse event tables with $\chi^2$ or Fisher test for comparision.

sstable manipulation

These function provides helper to create ss_table-esque output, with supporting function to convert them to flextable or huxtable.

• ssformat function to convert an arbitrary matrix to an sstable
• ss_flextable function to convert an sstable to a flextable
• ss_huxtable function to convert an sstable to a huxtable
• ft_sstheme function to decorate an arbitrary flextable to have an sstable-esque theme
• ht_theme_markdown and ht_theme_kable functions to decorate huxtable to follow a sstable-esque theme
• as_sstable converting objects to sstable

Plotting functions

• gg_boxcox same with MASS::boxcox but in ggplot
• ggsurvfit2 same as ggsurvfit::ggsurvfit but with strata separated. Note that it relies on tidy_survfit2 and does not work with ggsurvfit::survfit2 at the moment.
• gg_ajsurvplot plots one Aalen-Johansen curve for competing risks for the event of interest. Note that this is based on surviminer due to its flexibility
• gg_ajsurvplot2 plots two Aalen-Johansen curves for main risk (from bottom) and competing risk (from top). Note that this is based ggsurvfit with some limitations. However, it has the ability to return a dataset for more flexibility in plotting with ggplot.

OUCRU function written by Lam PK

• import.info and convert.info import and convert OUCRU dictionary to C306 style dict to use for data inspection
• insepct.data function to check data for error based on a dictionary by convert.info
• myformat.data function to reformat the data to readable form base on the dictionary by convert.info
• labAE function to determine a laboratory AE based on a threshold

Helper function written by Trinh Dong

• logist_summary based on a function by OUCRU Biostats group, reporting the OR of logistic regression
• subgroup_effect extends logist_summary to report the subgroup OR of variables that have interaction with a common covariate (i.e. treatment arm).
• mutate_f and summarise_f perform multiple assignment for data.frame by c(a,b):=list(v_a, v_b)
• simple_relevel relevels a factor correposnding to the level of another factor
• simple_recode pattern-based recoding of data by in-place replacement
• tidy_survfit2 same as \code{\link[ggsurvfit:tidy_survfit]{ggsurvfit:tidy_survfit}} but with strata saved for easier plotting. Note that it, at the moment, does not work with ggsurvfit::survfit2 ironically.

oucru-biostats/C306 documentation built on July 16, 2024, 2:33 p.m.