From 5a7b869170792e11fa11ab7a718b2466a9d29642 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Sat, 29 Apr 2023 07:04:09 +0200
Subject: [PATCH] Add a better management of and create translation exception
 when required for initiation codon

Former-commit-id: 878d919fdaad16e6e2645b62b3a53ef5d5e1ef2b
Former-commit-id: 3c3647cf114438a1ea9c3ff8c44e67e367929776
---
 detectors/cds/lib/toEmbl.awk    | 12 ++++++++++++
 detectors/cds/lib/translate.awk | 33 ++++++++++++++++++++++++++++-----
 2 files changed, 40 insertions(+), 5 deletions(-)

diff --git a/detectors/cds/lib/toEmbl.awk b/detectors/cds/lib/toEmbl.awk
index 01a9211..d26729b 100644
--- a/detectors/cds/lib/toEmbl.awk
+++ b/detectors/cds/lib/toEmbl.awk
@@ -96,7 +96,9 @@ function Unk(s) {
 /^c begin_entry/ {
 	Nexon = 0
 	FrameShift=0
+ 	nte = 0
 	delete Exon
+	delete TransExcep
 	next
 }
 
@@ -151,6 +153,12 @@ function Unk(s) {
   next
 }
 
+
+/^e trans_exception/ {
+  TransExcep[++nte] = "(pos:"$3",aa:"$4")"     # /transl_except=(pos:<location>,aa:<amino_acid>)
+  next
+}
+
 /^c end_entry/ {
 
 	GeneName = Unk(GeneName)
@@ -185,6 +193,10 @@ function Unk(s) {
 			QQualifier("pseudogene","unknown")
 			QQualifier("note","nonfunctional due to stop codon")
 		}
+
+		for (ie = 1 ; ie <= nte; ie++) {
+		   SQualifier("transl_except", TransExcep[ie])
+		}
 		
 		QQualifier("translation", Translat)
 	}
diff --git a/detectors/cds/lib/translate.awk b/detectors/cds/lib/translate.awk
index 4017f62..3a33beb 100644
--- a/detectors/cds/lib/translate.awk
+++ b/detectors/cds/lib/translate.awk
@@ -8,9 +8,16 @@ BEGIN {
 
 /^c end_entry/ {
   if (RevStrand) Cds = RevComplement(Cds)
-  Prot = Translate(substr(Cds, 1, length(Cds)-3))
-  if (Modif == "=")
-    Prot = "M" substr(Prot, 2)
+  Prot = Translate(substr(Cds, 1, length(Cds)-3),Modif,"chloroplast")
+
+  if (Modif == "=" && substr(Prot, 1,1) != "M") {
+    Prot = "M" substr(Prot, 2,length(Prot))
+    if (CdsStartPosFull==1) {
+      print "e trans_exception "CdsStartPos" Met"
+    } else {
+      print "e trans_exception " "ERROR" " Met"
+    }
+  }
   print "e translate " Prot
 }
 
@@ -21,19 +28,35 @@ BEGIN {
 /^c begin_entry/ {
   Cds = ""
   Iexon = 0
+  CdsStartPosFullv = 0
   next
 }
 
 /^e exon/ {
+  print "d cds " CDS
   RevStrand = ($6 == "-")
   if (++Iexon == 1) { # first is exon with start (even on - strand)
     Modif = $15
     gsub("\"", "", Modif)
     Modif = (RevStrand ? substr(Modif, 2, 1) : substr(Modif, 1, 1))
   }
-  if (RevStrand)
+  if (RevStrand){
     Cds = SubSeq(Seq, $3, $4) "" Cds
-  else
+    if (Iexon==1) {
+      if(($4 - $3 +1) >= 3) {
+        CdsStartPos="complement("$4-2".."$4")"
+        CdsStartPosFull=1
+      }
+    }}
+  else{
     Cds = Cds "" SubSeq(Seq, $3, $4)
+    if (Iexon==1) {
+      if (Iexon==1) {
+        if(($4 - $3 +1) >= 3) {
+          CdsStartPos=$3".."$3+2
+          CdsStartPosFull=1
+        }
+    }
+    }  }
   next
 }