Nana Mensah - 1st July 2018

This research benchmarks WGS antimicrobial resistance predictions, using ARIBA and the Resfinder database (v.3), with the gold standard broth microdilution method. Additionally, WGS predictions were evaluated in tandem with legacy methods (disk diffusion, agar dilution) that are routinely used due to their utility for AMR surveillance and clinical diagnostics.

The Makefile included contains instructions in the GNU make utility to run the analysis scripts in the correct order. With R and GNU make installed, call make from this directory on the command line to run the analysis (Linux tested only).

Warnings - As R performs type conversions throughout the scripts, many warnings are printed. An error from make as displayed below is the indicator that the script has not completed:

Makefile:17: recipe for target 'output' failed
make: *** [output] Error 1

• preprocessing/* - Scripts to create files for analysis. Output to data/clean/
• functions.R - All scripts import functions defined in this file
• tables.R - Generates data for tables in the manuscript
• create_db.R - Generates a csv that can be filtered to query individual isolate-test concordance results
• analysis.R - Performs statistical analyses on cleaned input data

• genotype.csv ; per-isolate gene presence/absence
• broth_ecoff.csv ; per-isolate broth microdilution results interpreted with ECOFFs
• broth_clinbp.csv ; per-isolate broth microdilution results interpreted with CBPs
• disk_clinbp.csv ; per-isolate disk diffusion results interpreted with CBPs
• agar_clinbp.csv ; per-isolate agar diliution results interpreted with CBPs

All outputs of this analysis can be found in the 'results/' and 'plots/' directories.

** Tested with R version 3.5.1 and RStudio 1.1.463 **