From b36ab2d29a3015b880be57705d9fff3c3b36c195 Mon Sep 17 00:00:00 2001
From: AndreaGuarracino <andreaguarracino@outlook.com>
Date: Wed, 9 Aug 2023 14:00:57 +0200
Subject: [PATCH] check PanSN in the input FASTA

---
 partition-before-pggb | 16 ++++++++++++++--
 pggb                  | 22 ++++++++++++++++++----
 2 files changed, 32 insertions(+), 6 deletions(-)

diff --git a/partition-before-pggb b/partition-before-pggb
index 1489f85..de252d3 100755
--- a/partition-before-pggb
+++ b/partition-before-pggb
@@ -157,12 +157,12 @@ fi
 # Mandatory parameters
 if [[ "$input_fasta" == false || $n_mappings == false ]]; then
     show_help=true
-    >&2 echo "ERROR: mandatory arguments -i and -n"
+    >&2 echo "[pggb] ERROR: mandatory arguments -i and -n"
 fi
 
 if (( "$n_mappings" < 2 )); then
     show_help=true
-    >&2 echo "ERROR: -n must be greater than or equal to 2"
+    >&2 echo "[pggb] ERROR: -n must be greater than or equal to 2"
 fi
 
 if [ $show_help == true ]; then
@@ -432,6 +432,18 @@ reporting:
   multiqc:            $multiqc
 EOT
 
+# Check Pangenome Sequence Naming (PanSN)
+if [ ! -f "${input_fasta}.fai" ]; then
+    echo "[pggb] ERROR: Index for $input_fasta does not exist. Please create it using 'samtools faidx $input_fasta'."
+    exit 1
+fi
+warning_emitted=0
+cut -f 1 "${input_fasta}.fai" | while read -r line; do
+  if [[ ! $line =~ ^([^#]+#)+[^#]+$ ]] && [[ $warning_emitted -eq 0 ]]; then
+      echo "[pggb] Warning: there are sequence names (like '$line') that do not match the Pangenome Sequence Naming (PanSN)."
+      warning_emitted=1
+  fi
+done
 
 #-------------------------------------------------------------------------------
 echo -e "\nRunning partitioning\n" >> "$log_file"
diff --git a/pggb b/pggb
index 821ee80..4ca3319 100755
--- a/pggb
+++ b/pggb
@@ -97,7 +97,7 @@ parse_numeric() {
       echo $value
       return 0
     else
-      echo "ERROR: Invalid input" >&2
+      echo "[pggb] ERROR: Invalid input" >&2
       return 1
     fi
   fi
@@ -106,7 +106,7 @@ parse_numeric() {
     M|m) value=$(echo "$value * 1000000" | bc) ;;
     G|g) value=$(echo "$value * 1000000000" | bc) ;;
     T|t) value=$(echo "$value * 1000000000000" | bc) ;;
-    *) echo "ERROR: Invalid suffix or unsupported suffix. Supported metric suffixes are k, K, m, M, g, G, t, T." >&2; return 1 ;;
+    *) echo "[pggb] ERROR: Invalid suffix or unsupported suffix. Supported metric suffixes are k, K, m, M, g, G, t, T." >&2; return 1 ;;
   esac
   printf "%.0f" $value
 }
@@ -180,12 +180,12 @@ fi
 # Mandatory parameters
 if [[ "$input_fasta" == false || $n_mappings == false ]]; then
     show_help=true
-    >&2 echo "ERROR: mandatory arguments -i and -n"
+    >&2 echo "[pggb] ERROR: mandatory arguments -i and -n"
 fi
 
 if (( "$n_mappings" < 2 )); then
     show_help=true
-    >&2 echo "ERROR: -n must be greater than or equal to 2"
+    >&2 echo "[pggb] ERROR: -n must be greater than or equal to 2"
 fi
 
 if [ $show_help == true ]; then
@@ -456,6 +456,20 @@ EOT
 
 
 echo -e "\nRunning pggb\n" >> "$log_file"
+
+# Check Pangenome Sequence Naming (PanSN)
+if [ ! -f "${input_fasta}.fai" ]; then
+    echo "[pggb] ERROR: Index for $input_fasta does not exist. Please create it using 'samtools faidx $input_fasta'."
+    exit 1
+fi
+warning_emitted=0
+cut -f 1 "${input_fasta}.fai" | while read -r line; do
+  if [[ ! $line =~ ^([^#]+#)+[^#]+$ ]] && [[ $warning_emitted -eq 0 ]]; then
+      echo "[pggb] Warning: there are sequence names (like '$line') that do not match the Pangenome Sequence Naming (PanSN)."
+      warning_emitted=1
+  fi
+done
+
 if [[ "$input_paf" == false ]]; then
   if [[ ! -s "$prefix_paf".alignments.$mapper.paf || $resume == false ]]; then