From f919a3152728396503c1d2ae9119b333e5f3d422 Mon Sep 17 00:00:00 2001
From: Jens Sauer <jsauer65@gmail.com>
Date: Mon, 16 Nov 2020 12:24:20 +0100
Subject: [PATCH] utils: Add sma_load_data()

This function loads the data and adjust some wrong column types. It
shoulb be used from all other scripts to load the data. It the same
source data for all following calculations.
The data is stored inside a 'tibble' from the 'tidyverse' package.
---
 utils.R | 46 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

diff --git a/utils.R b/utils.R
index 7d7bc48..c8c6105 100644
--- a/utils.R
+++ b/utils.R
@@ -3,6 +3,8 @@
 # License: GPL version 3
 # Jens Mathis Sauer (c) 2020
 
+library(tidyverse)
+
 #
 # Write plot to filename
 #
@@ -27,3 +29,47 @@ sma_plot_file <- function(fname, ftype, title_list, legend_list, fun, ...) {
 	# hide "null device 1" in garbage
 	garbage <- dev.off()
 }
+
+#
+# Load secMalASCT data
+#
+# This loads the secMalASCT data from a CSV file and adjusts some
+# columns to the correct type.
+#
+sma_load_data <- function(file) {
+	sma <- as_tibble(read.csv(file, header = TRUE))
+
+	# Type adjustments
+	sma$follow_up_multi <- as.logical(sma$follow_up_multi)
+	sma$nicotin_history <- as.logical(sma$nicotin_history)
+	sma$relapse_treatment_asct <- as.logical(sma$relapse_treatment_asct)
+	sma$relapse_treatment_chemotherapy <- as.logical(sma$relapse_treatment_chemotherapy)
+	sma$relapse_treatment_radiotherapy <- as.logical(sma$relapse_treatment_radiotherapy)
+	sma$relapse_treatment_surgery <- as.logical(sma$relapse_treatment_surgery)
+
+	sma$pre_rt <- as.logical(sma$pre_rt)
+	sma$pre_rt_type <- as.factor(sma$pre_rt_type)
+	sma$post_rt <- as.logical(sma$post_rt)
+	sma$post_rt_type <- as.factor(sma$post_rt_type)
+
+	sma$pre_alkylating_agents <- as.logical(sma$pre_alkylating_agents)
+	sma$pre_topoisomerase <- as.logical(sma$pre_topoisomerase)
+	sma$pre_anthracyclines <- as.logical(sma$pre_anthracyclines)
+
+	sma$peri_alkylating_agents <- as.logical(sma$peri_alkylating_agents)
+	sma$peri_topoisomerase <- as.logical(sma$peri_topoisomerase)
+	sma$peri_anthracyclines <- as.logical(sma$peri_anthracyclines)
+
+	sma$post_alkylating_agents <- as.logical(sma$post_alkylating_agents)
+	sma$post_topoisomerase <- as.logical(sma$post_topoisomerase)
+	sma$post_anthracyclines <- as.logical(sma$post_anthracyclines)
+
+	sma$maintenance_treatment <- as.logical(sma$maintenance_treatment)
+	sma$maintenance_treatment_time <- as.integer(sma$maintenance_treatment_time)
+	sma$maintenance_treatment_type <- as.factor(sma$maintenance_treatment_type)
+
+	sma$thalidomid <- as.logical(sma$thalidomid)
+	sma$bortezomib <- as.logical(sma$bortezomib)
+
+	return(sma)
+}