diff saint_wrapper.py @ 28:b7e5a9bfa620 draft

Uploaded

--- a/saint_wrapper.py	Thu Jan 28 16:31:30 2016 -0500
+++ b/saint_wrapper.py	Thu Jan 28 16:32:34 2016 -0500
@@ -1,10 +1,10 @@
 #######################################################################################
-# Python-code: SAINT wrapper 
+# Python-code: Dotplot Runner
 # Author: Adam L Borne
 # Contributers: Paul A Stewart, Brent Kuenzi
 #######################################################################################
-# This program takes a bait, prey and interactions file and uses SaintExpress to 
-# generate resulting matrices. This allows the software to run within galaxy. 
+# This runs SAINTexpress found at http://saint-apms.sourceforge.net/Main.html in
+# galaxy. 
 #######################################################################################
 # Copyright (C)  Adam Borne.
 # Permission is granted to copy, distribute and/or modify this document
@@ -16,18 +16,17 @@
 #######################################################################################
 ## REQUIRED INPUT ##
 
-# 1) inter_file: Interaction file listing experiment bait, assignement, gene name and 
-# spectrum count.
-# 2) prey_file: Prey file listing gene name, sequence legnth, and gene id.
-# 3) bait_file: Bait file listing bait experiment, assignment, and T or C. 
-# 4) vc_bool: Boolean used to determine if virtual controls are being used.
-# 5) vc_num: Number of virtual controls to be used. (default = 1)
-# 6) go_bool: Boolean used to determine used of external data set. 
-# 7) go_file: External data file if being used. (default = "None")
+# 1) list_file: SaintExpress output file.
+# 2) FDR1: Primary false discovery rate. (default = 0.01)
+# 3) FDR2: Secondary false discovery rate. (default = 0.025)
+# 4) spec_max: Maximum spectral count. (default = 50)
 #######################################################################################
+
+
 import os
 import sys
 
+
 inter_file = sys.argv[1]
 prey_file = sys.argv[2]
 bait_file = sys.argv[3]
@@ -39,54 +38,70 @@
 output_file = sys.argv[9]
 ins_path = sys.argv[10]
 
-# Verifies the install of SaintExpress and installs if not found.
-def first_run_check(): 
-	os.chdir(ins_path)
-	dirs_list = []
-	for (dirpath, dirnames, filename) in os.walk("./"):
-		dirs_list.extend(dirnames)
-		break
-	if r"SAINTexpress_v3.6.1__2015-05-03" in dirs_list:
-		pass
-	else: 
-		cmd = r"unzip SAINTexpress_v3.6.1__2015-05-03.zip" 
-		os.system(cmd)
-		os.chdir("./SAINTexpress_v3.6.1__2015-05-03")
-		cmd1 = r"make -j"
-		os.system(cmd1)
 
-def default_run(inter_file1,prey_file1,bait_file1,output_file1,num_of_rep1):
-	cmd = str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc " + r"-R" + str(num_of_rep1) + " " + str(inter_file1) + " " + str(prey_file1) + " " + str(bait_file1) 
-	os.system(cmd) 
-	open('list.txt')
-	os.rename('list.txt', str(output_file1)) 
+def first_run_check():
+    dirs_list = []
+    for (dirpath, dirnames, filename) in os.walk("./"):
+        dirs_list.extend(dirnames)
+        break
+    if r"SAINTexpress_v3.6.1__2015-05-03" in dirs_list:
+        pass
+    else:
+        cmd = r"unzip SAINTexpress_v3.6.1__2015-05-03.zip"
+        os.system(cmd)
+        os.chdir("./SAINTexpress_v3.6.1__2015-05-03")
+        cmd1 = r"make -j"
+        os.system(cmd1)
+
+
+def default_run(inter_file1, prey_file1, bait_file1, output_file1, num_of_rep1):
+    # Default is no virtual controls through purification and set replicates number.
+    cmd = (str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc " + r"-R"
+           + str(num_of_rep1) + " " + str(inter_file1) + " " + str(prey_file1) +
+           " " + str(bait_file1))
+    os.system(cmd) 
+    open('list.txt')
+    os.rename('list.txt', str(output_file1)) 
+
 
-def with_L(inter_file1,prey_file1,bait_file1,output_file1,vc_num1,num_of_rep1):
-	cmd = str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc "+ r"-R" + str(num_of_rep1) + " " + r"-L" + str(vc_num1) + " " + str(inter_file1) + " " + str(prey_file1) + " " + str(bait_file1) 
-	os.system(cmd) 
-	open('list.txt')
-	os.rename('list.txt', str(output_file1)) 
+def with_L(inter_file1, prey_file1, bait_file1, output_file1, vc_num1, num_of_rep1):
+    # L is the flag for Virtual Controls through Purification.
+    cmd = (str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc "+ r"-R"
+           + str(num_of_rep1) + " " + r"-L" + str(vc_num1) + " " + str(inter_file1) + " " +
+           str(prey_file1) + " " + str(bait_file1))
+    os.system(cmd) 
+    open('list.txt')
+    os.rename('list.txt', str(output_file1)) 
+
 
-def external_data_no_L(inter_file1,prey_file1,bait_file1,output_file1,go_file1,num_of_rep1):
-	cmd = str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc "+ r"-R" + str(num_of_rep1) + " " + str(inter_file1) + " " + str(prey_file1) + " " + str(bait_file1) + " " + str(go_file1)
-	os.system(cmd) 
-	open('list.txt')
-	os.rename('list.txt', str(output_file1)) 
+def external_data_no_L(inter_file1, prey_file1, bait_file1, output_file1, go_file1, num_of_rep1):
+    # Uses external data in the GO file format and no Virtual Controls.
+    cmd = (str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc "+ r"-R"
+           + str(num_of_rep1) + " " + str(inter_file1) + " " + str(prey_file1) + " " +
+           str(bait_file1) + " " + str(go_file1))
+    os.system(cmd) 
+    open('list.txt')
+    os.rename('list.txt', str(output_file1)) 
+
 
-def external_data_with_L(inter_file1,prey_file1,bait_file1,output_file1,go_file1,num_of_rep1,vc_num1):
-	cmd = str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc "+ r"-R" + str(num_of_rep1) + " " + r"-L" + str(vc_num1) + " " + str(inter_file1) + " " + str(prey_file1) + " " + str(bait_file1) + " " + str(go_file1)
-	os.system(cmd) 
-	open('list.txt')
-	os.rename('list.txt', str(output_file1)) 
+def external_data_with_L(inter_file1, prey_file1, bait_file1, output_file1, go_file1, num_of_rep1, vc_num1):
+    # Uses external data in the GO file format and Virtual Controls.
+    cmd = (str(ins_path) + r"/SAINTexpress_v3.6.1__2015-05-03/bin/SAINTexpress-spc "+ r"-R"
+           + str(num_of_rep1) + " " + r"-L" + str(vc_num1) + " " + str(inter_file1) + " " +
+           str(prey_file1) + " " + str(bait_file1) + " " + str(go_file1))
+    os.system(cmd) 
+    open('list.txt')
+    os.rename('list.txt', str(output_file1)) 
+
 
-first_run_check() 
-if (vc_bool == "true"):
-	if (go_bool == "false"):
-		with_L(inter_file, prey_file, bait_file, output_file, vc_num, num_of_rep)
-	elif (go_bool == "true"):
-		external_data_with_L(inter_file, prey_file, bait_file, output_file, go_file, num_of_rep, vc_num)
-elif (vc_bool == "false"):
-	if (go_bool == "false"):
-		default_run(inter_file, prey_file, bait_file, output_file, num_of_rep)
-	elif (go_bool == "true"):
-		external_data_no_L(inter_file, prey_file, bait_file, output_file, go_file, num_of_rep)
+first_run_check()
+if vc_bool == "true":
+    if go_bool == "false":
+        with_L(inter_file, prey_file, bait_file, output_file, vc_num, num_of_rep)
+    elif go_bool == "true":
+        external_data_with_L(inter_file, prey_file, bait_file, output_file, go_file, num_of_rep, vc_num)
+elif vc_bool == "false":
+    if go_bool == "false":
+        default_run(inter_file, prey_file, bait_file, output_file, num_of_rep)
+    elif go_bool == "true":
+        external_data_no_L(inter_file, prey_file, bait_file, output_file, go_file, num_of_rep)