From 2a826ea8ebb8fcbd43913e87cac8ebc3a131b4f0 Mon Sep 17 00:00:00 2001
From: "Vera N. Karlbauer" <vera@karlbauer.org>
Date: Sat, 4 Nov 2023 17:16:29 +0100
Subject: [PATCH] Function for calculating appropriate number of PCs to include
 for batch identification/correction

---
 data/Calculate_PC_Cutoff.R | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
 create mode 100644 data/Calculate_PC_Cutoff.R

diff --git a/data/Calculate_PC_Cutoff.R b/data/Calculate_PC_Cutoff.R
new file mode 100644
index 0000000..5a93226
--- /dev/null
+++ b/data/Calculate_PC_Cutoff.R
@@ -0,0 +1,14 @@
+### Function calculating number of PCs to be kept for batch identification and correction
+# threshold for N of PCs: start at PC1 and keep including PCs until drop in explained variance from PCi to PCi+2 < 1%
+# if threshold >2, include first 2 PCs (needed for plots)
+# include max 10 PCs
+
+pc_cutoff <- function(data){
+  for(i in 1:nrow(data)) {
+    if(i > 10) break
+    else if((data[i,]$var_explained - data[i + 2,]$var_explained) >=1) PC_number <- i 
+    else if(PC_number < 2) PC_number = 2
+    else break
+  }
+  return(PC_number)
+}