mirror of https://github.com/msberends/AMR
You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
125 lines
6.3 KiB
125 lines
6.3 KiB
# ==================================================================== # |
|
# TITLE # |
|
# Antimicrobial Resistance (AMR) Data Analysis for R # |
|
# # |
|
# SOURCE # |
|
# https://github.com/msberends/AMR # |
|
# # |
|
# LICENCE # |
|
# (c) 2018-2021 Berends MS, Luz CF et al. # |
|
# Developed at the University of Groningen, the Netherlands, in # |
|
# collaboration with non-profit organisations Certe Medical # |
|
# Diagnostics & Advice, and University Medical Center Groningen. # |
|
# # |
|
# This R package is free software; you can freely use and distribute # |
|
# it for both personal and commercial purposes under the terms of the # |
|
# GNU General Public License version 2.0 (GNU GPL-2), as published by # |
|
# the Free Software Foundation. # |
|
# We created this package for both routine data analysis and academic # |
|
# research and it was publicly released in the hope that it will be # |
|
# useful, but it comes WITHOUT ANY WARRANTY OR LIABILITY. # |
|
# # |
|
# Visit our website for the full manual and a complete tutorial about # |
|
# how to conduct AMR data analysis: https://msberends.github.io/AMR/ # |
|
# ==================================================================== # |
|
|
|
expect_equal(proportion_R(example_isolates$AMX), resistance(example_isolates$AMX)) |
|
expect_equal(proportion_SI(example_isolates$AMX), susceptibility(example_isolates$AMX)) |
|
# AMX resistance in `example_isolates` |
|
expect_equal(proportion_R(example_isolates$AMX), 0.5955556, tolerance = 0.0001) |
|
expect_equal(proportion_I(example_isolates$AMX), 0.002222222, tolerance = 0.0001) |
|
expect_equal(1 - proportion_R(example_isolates$AMX) - proportion_I(example_isolates$AMX), |
|
proportion_S(example_isolates$AMX)) |
|
expect_equal(proportion_R(example_isolates$AMX) + proportion_I(example_isolates$AMX), |
|
proportion_IR(example_isolates$AMX)) |
|
expect_equal(proportion_S(example_isolates$AMX) + proportion_I(example_isolates$AMX), |
|
proportion_SI(example_isolates$AMX)) |
|
|
|
if (AMR:::pkg_is_available("dplyr", min_version = "1.0.0")) { |
|
expect_equal(example_isolates %>% proportion_SI(AMC), |
|
0.7626397, |
|
tolerance = 0.0001) |
|
expect_equal(example_isolates %>% proportion_SI(AMC, GEN), |
|
0.9408, |
|
tolerance = 0.0001) |
|
expect_equal(example_isolates %>% proportion_SI(AMC, GEN, only_all_tested = TRUE), |
|
0.9382647, |
|
tolerance = 0.0001) |
|
|
|
# percentages |
|
expect_equal(example_isolates %>% |
|
group_by(hospital_id) %>% |
|
summarise(R = proportion_R(CIP, as_percent = TRUE), |
|
I = proportion_I(CIP, as_percent = TRUE), |
|
S = proportion_S(CIP, as_percent = TRUE), |
|
n = n_rsi(CIP), |
|
total = n()) %>% |
|
pull(n) %>% |
|
sum(), |
|
1409) |
|
|
|
# count of cases |
|
expect_equal(example_isolates %>% |
|
group_by(hospital_id) %>% |
|
summarise(cipro_p = proportion_SI(CIP, as_percent = TRUE), |
|
cipro_n = n_rsi(CIP), |
|
genta_p = proportion_SI(GEN, as_percent = TRUE), |
|
genta_n = n_rsi(GEN), |
|
combination_p = proportion_SI(CIP, GEN, as_percent = TRUE), |
|
combination_n = n_rsi(CIP, GEN)) %>% |
|
pull(combination_n), |
|
c(305, 617, 241, 711)) |
|
|
|
# proportion_df |
|
expect_equal( |
|
example_isolates %>% select(AMX) %>% proportion_df() %>% pull(value), |
|
c(example_isolates$AMX %>% proportion_SI(), |
|
example_isolates$AMX %>% proportion_R()) |
|
) |
|
expect_equal( |
|
example_isolates %>% select(AMX) %>% proportion_df(combine_IR = TRUE) %>% pull(value), |
|
c(example_isolates$AMX %>% proportion_S(), |
|
example_isolates$AMX %>% proportion_IR()) |
|
) |
|
expect_equal( |
|
example_isolates %>% select(AMX) %>% proportion_df(combine_SI = FALSE) %>% pull(value), |
|
c(example_isolates$AMX %>% proportion_S(), |
|
example_isolates$AMX %>% proportion_I(), |
|
example_isolates$AMX %>% proportion_R()) |
|
) |
|
} |
|
|
|
expect_warning(proportion_R(as.character(example_isolates$AMC))) |
|
expect_warning(proportion_S(as.character(example_isolates$AMC))) |
|
expect_warning(proportion_S(as.character(example_isolates$AMC, |
|
example_isolates$GEN))) |
|
|
|
expect_warning(n_rsi(as.character(example_isolates$AMC, |
|
example_isolates$GEN))) |
|
expect_equal(suppressWarnings(n_rsi(as.character(example_isolates$AMC, |
|
example_isolates$GEN))), |
|
1879) |
|
|
|
# check for errors |
|
expect_error(proportion_IR("test", minimum = "test")) |
|
expect_error(proportion_IR("test", as_percent = "test")) |
|
expect_error(proportion_I("test", minimum = "test")) |
|
expect_error(proportion_I("test", as_percent = "test")) |
|
expect_error(proportion_S("test", minimum = "test")) |
|
expect_error(proportion_S("test", as_percent = "test")) |
|
expect_error(proportion_S("test", also_single_tested = TRUE)) |
|
|
|
# check too low amount of isolates |
|
expect_identical(suppressWarnings(proportion_R(example_isolates$AMX, minimum = nrow(example_isolates) + 1)), |
|
NA_real_) |
|
expect_identical(suppressWarnings(proportion_I(example_isolates$AMX, minimum = nrow(example_isolates) + 1)), |
|
NA_real_) |
|
expect_identical(suppressWarnings(proportion_S(example_isolates$AMX, minimum = nrow(example_isolates) + 1)), |
|
NA_real_) |
|
|
|
# warning for speed loss |
|
expect_warning(proportion_R(as.character(example_isolates$GEN))) |
|
expect_warning(proportion_I(as.character(example_isolates$GEN))) |
|
expect_warning(proportion_S(example_isolates$AMC, as.character(example_isolates$GEN))) |
|
expect_error(proportion_df(c("A", "B", "C"))) |
|
expect_error(proportion_df(example_isolates[, "date"]))
|
|
|