From bd040a279b80b1f68e9bfe5401dd2a0889b7cb25 Mon Sep 17 00:00:00 2001
From: anastasiia <anastasiia.petrova@mpi-bn.mpg.de>
Date: Thu, 15 Nov 2018 14:13:08 +0100
Subject: [PATCH] writing the output file

---
 find_exons.py | 44 ++++++++++++++++++++++++++++++++++----------
 1 file changed, 34 insertions(+), 10 deletions(-)

diff --git a/find_exons.py b/find_exons.py
index 3aec43f..43874b1 100644
--- a/find_exons.py
+++ b/find_exons.py
@@ -69,6 +69,8 @@ def get_name_from_path(full_path):
 
 def check_existing_input_files(args):
 
+	print('checking the input files')
+
 	if not os.path.isfile(args.gtf_genes):
 		print('please make sure the .gtf file with genes exists')
 		sys.exit()
@@ -106,6 +108,8 @@ def check_existing_input_files(args):
 
 def procede_gtf_files(gtf_genes, gtf_exons, genes_of_interest):
 
+	logger.info("looking for genes of interest in gtf files")
+
 	all_genes = {}
 
 	for gene_name in genes_of_interest:
@@ -151,11 +155,6 @@ def procede_gtf_files(gtf_genes, gtf_exons, genes_of_interest):
 			if gene_full_name in all_genes.keys():
 				all_genes[gene_full_name]['exons'].append([exon_length, exon_number, tsl])
 
-	"""
-	for gene in all_genes:
-		print(gene, all_genes[gene])
-		print()
-	"""
 	return all_genes
 
 def make_plot(x_array, y_array, gene_names, output_directory, figure_name, color_name):
@@ -179,6 +178,13 @@ def make_plot(x_array, y_array, gene_names, output_directory, figure_name, color
 
 def plot_and_output(all_genes, output_directory):
 
+	logger.info("preparing for plotting")
+
+	gene_names = []
+	gene_lengths = []
+	exons_lengths = []
+	scores = []
+
 	#find the sum length of exons for each gene
 	for gene in all_genes:
 		#first sort the exons array
@@ -187,17 +193,35 @@ def plot_and_output(all_genes, output_directory):
 
 		check_exon = 1
 		sum_length = 0
-		last_exon = sorted_exons[-1][1]
+		last_exon = sorted_exons[-1][1] 
 
 		while check_exon <= last_exon:
 			if sorted_exons[0][1] == check_exon:
-				print(check_exon)
 				sum_length = sum_length + sorted_exons[0][0]
-				print(sum_length)
-				sorted_exons = [x for x in sorted_exons if x[1] != check_exon]
-				print(sorted_exons)
+				sorted_exons = [x for x in sorted_exons if x[1] != check_exon] #cut the current sorted_exons array so that there are no occurences of the current exon there
 			check_exon = check_exon + 1
 
+		score = "%.2f" % round((sum_length * 100)/all_genes[gene]['gene_length'], 2)
+
+		gene_names.append(gene)
+		gene_lengths.append(all_genes[gene]['gene_length'])
+		exons_lengths.append(sum_length)
+		scores.append(score)
+
+	logger.info("writing the output file")
+	
+	#now write an output file containing names of genes, their length, the sum length of exons and the prozent of exons in the gene length as score
+	output_file = open(os.path.join(output_directory, "genes_exons_correlation.txt"), 'w')
+	header = ["gene_name", "exons_len_percentage", "gene_length", "exons_sum_length"]
+	output_file.write('\t'.join(header) + '\n') #write the header
+
+	for i in range(len(gene_names)):
+		output_file.write('\t'.join([gene_names[i], str(scores[i]), str(gene_lengths[i]), str(exons_lengths[i])]) + '\n')
+
+	output_file.close()
+
+
+
 
 def main():