From d901750fa567b72fdce7656a11aca29cc6079d9b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Karatu=C4=9F=20Ozan=20Bircan?= <karatugo@ebi.ac.uk>
Date: Thu, 25 Jul 2024 17:16:56 +0100
Subject: [PATCH] feat: Split .tsv.gz files into two

---
 misc/split_files.sh | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 misc/split_files.sh

diff --git a/misc/split_files.sh b/misc/split_files.sh
new file mode 100644
index 0000000..cd7579c
--- /dev/null
+++ b/misc/split_files.sh
@@ -0,0 +1,33 @@
+#!/bin/bash
+#SBATCH --job-name=split
+#SBATCH --time=01:00:00
+#SBATCH --mem=4G
+#SBATCH --cpus-per-task=1
+#SBATCH --partition=datamover
+#SBATCH --mail-type=ALL
+#SBATCH --mail-user=karatugo@ebi.ac.uk
+
+# TODO: Define the full file paths
+INPUT_FILE=""
+OUTPUT_FILE_1=""
+OUTPUT_FILE_2=""
+
+# Extract the header
+HEADER=$(zcat "$INPUT_FILE" | head -n 1)
+
+# Count the total number of lines in the file
+TOTAL_LINES=$(zcat "$INPUT_FILE" | wc -l)
+
+# Calculate the midpoint
+MIDPOINT=$(( (TOTAL_LINES - 1) / 2 ))
+
+# Split the file into two parts
+zcat "$INPUT_FILE" | head -n $((MIDPOINT + 1)) | gzip > "$OUTPUT_FILE_1"
+{
+  echo "$HEADER"
+  zcat "$INPUT_FILE" | tail -n +$((MIDPOINT + 2))
+} | gzip > "$OUTPUT_FILE_2"
+
+echo "File has been split into two parts:"
+echo "First part: $OUTPUT_FILE_1"
+echo "Second part: $OUTPUT_FILE_2"