From 384dd721255af7b900225fb3eedf2c51163da179 Mon Sep 17 00:00:00 2001
From: Auden Cote-L'Heureux <52716489+AudenCote@users.noreply.github.com>
Date: Fri, 16 Feb 2024 14:06:06 -0500
Subject: [PATCH] Adding check to 4_InFrameStopCodonEstimator.py

---
 PTL1/Transcriptomes/Scripts/4_InFrameStopCodonEstimator.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/PTL1/Transcriptomes/Scripts/4_InFrameStopCodonEstimator.py b/PTL1/Transcriptomes/Scripts/4_InFrameStopCodonEstimator.py
index ae21d52..fa2b260 100644
--- a/PTL1/Transcriptomes/Scripts/4_InFrameStopCodonEstimator.py
+++ b/PTL1/Transcriptomes/Scripts/4_InFrameStopCodonEstimator.py
@@ -275,7 +275,6 @@ def prep_translations(args):
 	inFasta = [i for i in SeqIO.parse(args.input_file,'fasta')]
 
 	prot_dict = {}
-
 	for i in intsv:
 #		print i
 		prot_dict.setdefault(i.split('\t')[0],[])
@@ -294,6 +293,10 @@ def prep_translations(args):
 			else:
 				prot_dict[i.split('\t')[0]].append(int(i.split('\t')[7])-4)
 
+	if len(list(prot_dict.keys())) < 50:
+		print('\nStop codon estimation CANCELLED for taxon ' + args.input_file[:10] + ' because its file contains fewer than 50 sequences. This check occurred in script 4.\n')
+		exit()
+
 
 	#------------- Prep translation with 'TAA' as the only Stop -------------#