Not all msa print

src/total_data_format.py

@@ -83,7 +83,7 @@ last_threshold=threshold
 data={}
 bases=[]
 reads=[]
-MSA=[]
+MSA={}
 for threshold in files :
 	data["order"]=[]
 	for exp_name in files[threshold] :
@@ -94,6 +94,7 @@ for threshold in files :
 		header_size=0
 		for dir in files[threshold][exp_name]:
 			data[exp_name][dir] = {}
+			MSA[exp_name]=[]
 			data[exp_name]["order"]=data[exp_name]["order"] + [dir]
 			name_file = os.path.join(EXP,exp_name,dir,RESULT_FOLDER,files[threshold][exp_name][dir])
 			try:
@@ -115,8 +116,9 @@ for threshold in files :
 							bases.append(base_pair)
 						if (nb_read not in reads):
 							reads.append(nb_read)
-						if (msa not in MSA):
-							MSA.append(msa)
+
+						if (msa not in MSA[exp_name]):
+							MSA[exp_name].append(msa)
 
 						data[exp_name][dir][base_pair][nb_read][msa] = {}
 						for i in range(3,header_size):
@@ -125,6 +127,7 @@ for threshold in files :
 						header=tab_line
 						header_size=len(header)
 				read.close()
+				MSA[exp_name]=sorted(MSA[exp_name])
 			except:
 				print(name_file , "don't exists.")
 
@@ -157,12 +160,12 @@ for threshold in files :
 			for dir in data[exp_name]["order"]:
 				if dir == "":
 					output_mean.write("," + exp_name)
-					for msa in MSA:
+					for msa in MSA[exp_name]:
 						output_mean.write(",")
 				else:
 					if(not re.search("^sd_",attribute)):
 						output_all.write("," + os.path.join(exp_name,dir))
-						for msa in MSA:
+						for msa in MSA[exp_name]:
 							output_all.write(",")
 		output_mean.write("\nlenght,cover,")
 		if(not re.search("^sd_",attribute)):
@@ -170,12 +173,12 @@ for threshold in files :
 		for exp_name in data["order"]:
 			for dir in data[exp_name]["order"]:
 				if dir == "":
-					for msa in MSA:
+					for msa in MSA[exp_name]:
 						output_mean.write("," + msa)
 					output_mean.write(",")
 				else:
 					if(not re.search("^sd_",attribute)):
-						for msa in MSA:
+						for msa in MSA[exp_name]:
 							output_all.write("," + msa)
 						output_all.write(",")
 		output_mean.write("\n")
@@ -184,7 +187,6 @@ for threshold in files :
 
 		# Writing data
 		bases=sorted(bases)
-		MSA=sorted(MSA)
 		reads=sorted(reads)
 		for base in bases :
 			for read in reads :
@@ -194,7 +196,7 @@ for threshold in files :
 				for exp_name in data["order"]:
 					for dir in data[exp_name]["order"]:
 						if dir == "":
-							for msa in MSA:
+							for msa in MSA[exp_name]:
 								if (base in data[exp_name][dir]) and (read in data[exp_name][dir][base]) and (msa in data[exp_name][dir][base][read]):
 									output_mean.write("," + data[exp_name][dir][base][read][msa][attribute])
 								else:
@@ -202,7 +204,7 @@ for threshold in files :
 							output_mean.write(",")
 						else:
 							if(not re.search("^sd_",attribute)):
-								for msa in MSA:
+								for msa in MSA[exp_name]:
 									if (base in data[exp_name][dir]) and (read in data[exp_name][dir][base]) and (msa in data[exp_name][dir][base][read]):
 										output_all.write("," + data[exp_name][dir][base][read][msa][attribute])
 									else: