generated from opensafely/research-template
/
venn_diagram.R
245 lines (189 loc) · 9.17 KB
/
venn_diagram.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
## =============================================================================
## Purpose: Create venn diagrams
##
## Author: Yinghui Wei
##
## Reviewer: Renin Toms, Venexia Walker
##
## Date: 6 December 2021; updated 10 January 2022; updated 27 January 2022
##
## Data: Post covid vaccinated project study population
##
## Content: to create a Venn diagram for each outcome outlining overlap in
## reporting from different data sources
## Output: Venn diagrams in SVG files, venn_diagram_number_check.csv
## =============================================================================
library(data.table)
library(readr)
library(dplyr)
library(purrr)
args <- commandArgs(trailingOnly=TRUE)
if(length(args)==0){
# use for interactive testing
cohort_name <- ""
} else {
cohort_name <- args[[1]]
}
fs::dir_create(here::here("output", "not-for-review"))
fs::dir_create(here::here("output", "review", "venn-diagrams"))
venn_output <- function(group){
# Identify active outcomes ---------------------------------------------------
active_analyses <- readr::read_rds("lib/active_analyses.rds")
# added extra statement to include only those with venn == TRUE - because some diabetes outcomes only use one data source and so venn is not applicable
outcomes <- active_analyses[active_analyses$active==TRUE & active_analyses$venn==TRUE & active_analyses$outcome_group==group,]$outcome_variable
if(length(outcomes) == 0){
print(paste0("No venn diagram generated for outcome group ",group))
} else{
# Load data ------------------------------------------------------------------
input <- readr::read_rds(paste0("output/venn.rds"))
end_dates <- read_rds(paste0("output/follow_up_end_dates_",group,".rds"))
input_stage1 <- readr::read_rds(paste0("output/input_stage1_",group,".rds"))
input_stage1 <- input_stage1[input_stage1$sub_bin_covid19_confirmed_history==FALSE,]
input <- input[input$patient_id %in% input_stage1$patient_id,]
input<- input %>% left_join(end_dates, by="patient_id")
rm(input_stage1,end_dates)
# Create empty table ---------------------------------------------------------
df <- data.frame(outcome = character(),
only_snomed = numeric(),
only_hes = numeric(),
only_death = numeric(),
snomed_hes = numeric(),
snomed_death = numeric(),
hes_death = numeric(),
snomed_hes_death = numeric(),
total_snomed = numeric(),
total_hes = numeric(),
total_death = numeric(),
total = numeric(),
stringsAsFactors = FALSE)
# Populate table and make Venn for each outcome ------------------------------
for (outcome in outcomes) {
print(paste0("Working on ", outcome))
# Restrict data to that relevant to the given outcome ----------------------
tmp <- input[!is.na(input[,outcome]),c("patient_id","index_date",paste0(gsub("out_date_","", outcome),"_follow_up_end"), colnames(input)[grepl(outcome,colnames(input))])]
colnames(tmp) <- gsub(paste0("tmp_",outcome,"_"),"",colnames(tmp))
setnames(tmp,
old=c(paste0(gsub("out_date_","", outcome),"_follow_up_end"),
outcome),
new=c("follow_up_end",
"event_date"))
tmp <- tmp %>% filter(follow_up_end >= index_date)
# Impose follow-up start and end dates on events dates
event_cols <- c("snomed","hes","death","event_date")
for(colname in event_cols){
if(colname %in% colnames(tmp)){
tmp <- tmp %>% mutate(!!sym(colname) := replace(!!sym(colname), which(!!sym(colname)>follow_up_end | !!sym(colname)<index_date), NA))
}
}
# Identify and add missing columns -----------------------------------------
complete <- data.frame(patient_id = tmp$patient_id,
snomed = as.Date(NA),
hes = as.Date(NA),
death = as.Date(NA))
#colnames(complete) <- c("patient_id",paste0("tmp_",outcome,c("_snomed","_hes","_death")))
complete[,setdiff(colnames(tmp),"patient_id")] <- NULL
notused <- NULL
if (ncol(complete)>1) {
tmp <- merge(tmp, complete, by = c("patient_id"))
notused <- setdiff(colnames(complete),"patient_id")
}
# Calculate the number contributing to each source combo -------------------
tmp$snomed_contributing <- !is.na(tmp$snomed) &
is.na(tmp$hes) &
is.na(tmp$death)
tmp$hes_contributing <- is.na(tmp$snomed) &
!is.na(tmp$hes) &
is.na(tmp$death)
tmp$death_contributing <- is.na(tmp$snomed) &
is.na(tmp$hes) &
!is.na(tmp$death)
tmp$snomed_hes_contributing <- !is.na(tmp$snomed) &
!is.na(tmp$hes) &
is.na(tmp$death)
tmp$hes_death_contributing <- is.na(tmp$snomed) &
!is.na(tmp$hes) &
!is.na(tmp$death)
tmp$snomed_death_contributing <- !is.na(tmp$snomed) &
is.na(tmp$hes) &
!is.na(tmp$death)
tmp$snomed_hes_death_contributing <- !is.na(tmp$snomed) &
!is.na(tmp$hes) &
!is.na(tmp$death)
df[nrow(df)+1,] <- c(outcome,
only_snomed = nrow(tmp %>% filter(snomed_contributing==T)),
only_hes = nrow(tmp %>% filter(hes_contributing==T)),
only_death = nrow(tmp %>% filter(death_contributing==T)),
snomed_hes = nrow(tmp %>% filter(snomed_hes_contributing==T)),
snomed_death = nrow(tmp %>% filter(snomed_death_contributing==T)),
hes_death = nrow(tmp %>% filter(hes_death_contributing==T)),
snomed_hes_death = nrow(tmp %>% filter(snomed_hes_death_contributing==T)),
total_snomed = nrow(tmp %>% filter(!is.na(snomed))),
total_hes = nrow(tmp %>% filter(!is.na(hes))),
total_death = nrow(tmp %>% filter(!is.na(death))),
total = nrow(tmp %>% filter(!is.na(event_date))))
# Remove sources not in study definition from Venn plots and summary -------
source_combos <- c("only_snomed","only_hes","only_death","snomed_hes","snomed_death","hes_death","snomed_hes_death")
source_consid <- source_combos
if (!is.null(notused)) {
for (i in notused) {
# Add variables to consider for Venn plot to vector
source_consid <- source_combos[!grepl(i,source_combos)]
# Replace unused sources with NA in summary table
for (j in setdiff(source_combos,source_consid)) {
df[df$outcome==outcome,j] <- NA
}
}
}
# Proceed to create Venn diagram if all source combos exceed 5 -------------
# if (min(as.numeric(df[df$outcome==outcome,source_consid]))>5) {
# Calculate contents of each Venn cell for plotting ----------------------
index1 <- integer(0)
index2 <- integer(0)
index3 <- integer(0)
if ("only_snomed" %in% source_consid) {
index1 <- which(!is.na(tmp$snomed))
}
if ("only_hes" %in% source_consid) {
index2 <- which(!is.na(tmp$hes))
}
if ("only_death" %in% source_consid) {
index3 <- which(!is.na(tmp$death))
}
index <- list(index1, index2, index3)
names(index) <- c("Primary care", "Secondary care", "Death record")
index <- Filter(length, index)
# Fix colours --------------------------------------------------------------
mycol <- c(ifelse("Primary care" %in% names(index),"thistle",""),
ifelse("Secondary care" %in% names(index),"lightcyan",""),
ifelse("Death record" %in% names(index),"lemonchiffon",""))
mycol <- mycol[mycol!=""]
# Make Venn diagram --------------------------------------------------------
# only make venn diagram if list is not of length 0
if(length(index) == 0){
print("outcome venn not constructed - empty list")
} else{
svglite::svglite(file = paste0("output/review/venn-diagrams/venn_diagram_",group,"_",gsub("out_date_","",outcome),".svg"))
g <- ggvenn::ggvenn(
index,
fill_color = mycol,
stroke_color = "white",
text_size = 5,
set_name_size = 5,
fill_alpha = 0.9
) + ggplot2::ggtitle(active_analyses[active_analyses$outcome_variable==outcome,]$outcome) +
ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5, size = 15, face = "bold"))
print(g)
dev.off()
}
}
# Save summary file ----------------------------------------------------------
write.csv(df, file = paste0("output/review/venn-diagrams/venn_diagram_number_check_", group,".csv"), row.names = F)
}
}
# Run function using specified commandArgs and different outcome groups
active_analyses <- read_rds("lib/active_analyses.rds")
active_analyses <- active_analyses %>% filter(active==TRUE)
groups <- unique(active_analyses$outcome_group)
for(i in groups){
venn_output(i)
}