This repo contains the tool for looking at MICAG, as well as all the code to reproduce the analysis in the Data Challenge report. Note: the original data is not present in the repo and must be accessed through Vivli: https://vivli.org/
Main scripts are:
DATA CHALLENGE SPECIFIC
- 0_preparation_data_cleaning.R : This cleans and synthesises the datasets used in the Data Challenge. Must be run first in order to recreate the analysis.
GENERALISABLE TOOL (Note: these must be run in the order 1a, 1b, 2, 3.)
- 1a_TOOL_screening_plots.R : Creates MIC distribution plots by Group for specified characteristics and bacteria/antibiotic combinations.
- 1b_TOOL_screening_plots_time.R : As above, but includes the extra dimension of time.
- 2_TOOL_regression.R : Runs the regresssion analysis to check for confounders on the specified bacteria/antibiotic combination.
- 3_TOOL_index_analysis : Creates plots with the index analysis for all bacteria/antibiotics included in the dataset.
DATA CHALLENGE SPECIFIC
- 4a_samples_generation.R : this file creates the input data for Figure 4B in the report.
- 4b_Figures_data_challenge.R : This recreates the specific figures for our data challenge entry.
Other files in the repo are:
- Folder initial_testing_not_used : These are scripts written in the process of developing the tool, but are no longer used.
- income.csv : this contains mapping from the countries to World Bank income group.
- who-regions.csv : this contains the mapping from countries to WHO regions.
The TOOL can also be used with other data. In order to do this a cleaned dataset must be provided in a subfolder: data/full_data.csv Columns must include: gender, organism, antibiotic, mic, and any other characteristics the user wants to split the data by. The mic column must be numeric, and the other columns either character or factor. There may be some ggplot warnings if not all the antibiotic-bacteria combinations have all the groupings, but these can be ignored.
Packages required are: data.table, cowplot, ggplot2, MASS, tidyverse, patchwork.