annotate SAINT_preprocessing.py @ 4:019e60bd3f7f draft

Uploaded
author bornea
date Tue, 15 Mar 2016 15:59:24 -0400
parents
children 1b0547d3c7bc
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
1 #######################################################################################
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
2 # Python-code: SAINT pre-processing from Scaffold "Samples Report" output
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
3 # Author: Brent Kuenzi
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
4 #######################################################################################
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
5 # This program reads in a raw Scaffold "Samples Report" output and a user generated
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
6 # bait file and autoformats it into prey and interaction files for SAINTexpress
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
7 # analysis
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
8 #######################################################################################
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
9 # Copyright (C) Brent Kuenzi.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
10 # Permission is granted to copy, distribute and/or modify this document
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
11 # under the terms of the GNU Free Documentation License, Version 1.3
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
12 # or any later version published by the Free Software Foundation;
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
13 # with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
14 # A copy of the license is included in the section entitled "GNU
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
15 # Free Documentation License".
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
16 #######################################################################################
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
17 ## REQUIRED INPUT ##
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
18
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
19 # 1) infile: Scaffold "Samples Report" output
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
20 # 2) baitfile: SAINT formatted bait file generated in Galaxy
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
21 # 3) fasta_db: fasta database for use (defaults to SwissProt_HUMAN_2014_08.fasta)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
22 # 4) prey: Y or N for generating a prey file
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
23 # 5) make_bait: String of bait names, assignment, and test or control boolean
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
24 #######################################################################################
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
25
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
26 import sys
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
27 import os.path
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
28
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
29
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
30 infile = sys.argv[1]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
31 #Scaffold "Samples Report" output.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
32 prey = sys.argv[2]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
33 # Y or N boolean from Galaxy.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
34 fasta_db = sys.argv[3]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
35 tool_path = sys.argv[8]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
36 if fasta_db == "None":
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
37 fasta_db = str(tool_path) + "/SwissProt_HUMAN_2014_08.fasta"
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
38 make_bait = sys.argv[6]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
39 bait_bool = sys.argv[9]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
40
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
41
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
42 def bait_create(baits, infile):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
43 # Verifies the Baits are valid in the Scaffold file and writes the Bait.txt.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
44 baits = make_bait.split()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
45 i = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
46 bait_file_tmp = open("bait.txt", "w")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
47 order = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
48 bait_cache = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
49 while i < len(baits):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
50 if baits[i+2] == "true":
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
51 T_C = "C"
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
52 else:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
53 T_C = "T"
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
54 bait_line = baits[i] + "\t" + baits[i+1] + "\t" + T_C + "\n"
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
55 read_infile = open(infile, "r")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
56 for input_line in read_infile:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
57 input_line = input_line.strip()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
58 temp = input_line.split('\t')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
59 if "Quantitative Variance" in str(temp):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
60 if baits[i] in temp:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
61 number_bait = temp.index(str(baits[i]))
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
62 number_bait = number_bait - 9
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
63 bait_cache.append((number_bait, str(bait_line)))
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
64 # Locates the Bait names in the column names and then sets the Baits in the
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
65 # correct order in the cache thus the - 9 because the baits start at the 9th
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
66 # column.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
67 else:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
68 print "Error: bad bait " + str(baits[i])
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
69 sys.exit()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
70 else:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
71 pass
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
72 i = i + 3
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
73
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
74 bait_cache.sort()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
75 for cache_line in bait_cache:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
76 bait_file_tmp.write(cache_line[1])
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
77
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
78 bait_file_tmp.close()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
79
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
80 if bait_bool == 'false':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
81 bait_create(make_bait, infile)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
82 baitfile = "bait.txt"
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
83 else:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
84 bait_temp_file = open(sys.argv[10], 'r')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
85 bait_cache = bait_temp_file.readlines()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
86 bait_file_tmp = open("bait.txt", "wr")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
87 for cache_line in bait_cache:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
88 bait_file_tmp.write(cache_line)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
89 bait_file_tmp.close()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
90 baitfile = "bait.txt"
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
91
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
92
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
93 class ReturnValue1(object):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
94 def __init__(self, sequence, gene):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
95 self.seqlength = sequence
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
96 self.genename = gene
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
97
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
98
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
99 class ReturnValue2(object):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
100 def __init__(self, getdata, getproteins, getheader):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
101 self.data = getdata
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
102 self.proteins = getproteins
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
103 self.header = getheader
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
104
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
105
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
106 def main(Scaffold_input, baits):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
107 bait_check(baitfile, Scaffold_input)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
108 make_inter(Scaffold_input)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
109 if prey == 'true':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
110 make_prey(Scaffold_input)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
111 no_error_inter(Scaffold_input)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
112 os.rename('prey.txt', sys.argv[5])
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
113 elif prey == 'false':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
114 if os.path.isfile('error proteins.txt') == True:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
115 no_error_inter(Scaffold_input)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
116 pass
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
117 elif prey != 'true' or 'false':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
118 sys.exit("Invalid Prey Argument: Y or N")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
119
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
120
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
121 def get_info(uniprot_accession_in):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
122 # Get aminoacid lengths and gene name.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
123 error = open('error proteins.txt', 'a+')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
124 data = open(fasta_db, 'r')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
125 data_lines = data.readlines()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
126 db_len = len(data_lines)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
127 seqlength = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
128 count = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
129 for data_line in data_lines:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
130 if ">sp" in data_line:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
131 namer = data_line.split("|")[2]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
132 if uniprot_accession_in == data_line.split("|")[1]:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
133 match = count + 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
134 if 'GN=' in data_line:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
135 lst = data_linedata_line.split('GN=')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
136 lst2 = lst[1].split(' ')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
137 genename = lst2[0]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
138 if 'GN=' not in data_line:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
139 genename = 'NA'
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
140 while ">sp" not in data_lines[match]:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
141 if match <= db_len:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
142 seqlength = seqlength + len(data_lines[match].strip())
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
143 match = match + 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
144 else:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
145 break
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
146 return ReturnValue1(seqlength, genename)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
147 elif uniprot_accession_in == namer.split(" ")[0]:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
148 match = count + 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
149 # Ensures consistent spacing throughout.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
150 if 'GN=' in data_line:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
151 lst = data_line.split('GN=')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
152 lst2 = lst[1].split(' ')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
153 genename = lst2[0]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
154 if 'GN=' not in data_line:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
155 genename = 'NA'
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
156 while ">sp" not in data_lines[match]:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
157 if match <= db_len:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
158 seqlength = seqlength + len(data_lines[match].strip())
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
159 match = match + 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
160 else:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
161 break
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
162 return ReturnValue1(seqlength, genename)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
163 count = count + 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
164 if seqlength == 0:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
165 error.write(uniprot_accession_in + '\t' + "Uniprot not in Fasta" + '\n')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
166 error.close
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
167 seqlength = 'NA'
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
168 genename = 'NA'
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
169 return ReturnValue1(seqlength, genename)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
170
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
171
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
172 def readtab(infile):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
173 with open(infile, 'r') as input_file:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
174 # read in tab-delim text
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
175 output = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
176 for input_line in input_file:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
177 input_line = input_line.strip()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
178 temp = input_line.split('\t')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
179 output.append(temp)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
180 return output
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
181
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
182
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
183 def read_Scaffold(Scaffold_input):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
184 # Get data, proteins and header from Scaffold output
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
185 dupes = readtab(Scaffold_input)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
186 cnt = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
187 for Scaffold_line in dupes:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
188 cnt += 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
189 if Scaffold_line[0] == '#':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
190 # Finds the start of second header.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
191 header_start = cnt-1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
192 header = dupes[header_start]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
193 prot_start = header.index("Accession Number")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
194 data = dupes[header_start+1:len(dupes)-2]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
195 # Cut off blank line and END OF FILE.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
196 proteins = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
197 for Scaffold_line in data:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
198 Scaffold_line[4] = Scaffold_line[4].split()[0]
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
199 # Removes the (+##) that sometimes is attached.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
200 for protein in data:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
201 proteins.append(protein[prot_start])
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
202 return ReturnValue2(data, proteins, header)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
203
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
204
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
205 def make_inter(Scaffold_input):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
206 bait = readtab(baitfile)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
207 data = read_Scaffold(Scaffold_input).data
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
208 header = read_Scaffold(Scaffold_input).header
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
209 proteins = read_Scaffold(Scaffold_input).proteins
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
210 bait_index = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
211 for bait_line in bait:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
212 bait_index.append(header.index(bait_line[0]))
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
213 # Find just the baits defined in bait file.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
214 with open('inter.txt', 'w') as inter_file:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
215 a = 0; l = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
216 for bb in bait:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
217 for lst in data:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
218 inter_file.write(header[bait_index[l]] + '\t' + bb[1] + '\t' + proteins[a] + '\t'
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
219 + lst[bait_index[l]] + '\n')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
220 a += 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
221 if a == len(proteins):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
222 a = 0; l += 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
223
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
224
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
225 def make_prey(Scaffold_input):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
226 proteins = read_Scaffold(Scaffold_input).proteins
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
227 output_file = open("prey.txt", 'w')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
228 for protein in proteins:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
229 protein = protein.replace("\n", "")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
230 # Remove \n for input into function.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
231 protein = protein.replace("\r", "")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
232 # Ditto for \r.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
233 seq = get_info(protein).seqlength
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
234 GN = get_info(protein).genename
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
235 if seq != 'NA':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
236 output_file.write(protein + "\t" + str(seq) + "\t" + str(GN) + "\n")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
237 output_file.close()
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
238
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
239
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
240 def no_error_inter(Scaffold_input):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
241 # Remake inter file without protein errors from Uniprot.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
242 err = readtab("error proteins.txt")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
243 bait = readtab(baitfile)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
244 data = read_Scaffold(Scaffold_input).data
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
245 header = read_Scaffold(Scaffold_input).header
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
246 bait_index = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
247 for bait_line in bait:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
248 bait_index.append(header.index(bait_line[0]))
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
249 proteins = read_Scaffold(Scaffold_input).proteins
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
250 errors = []
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
251 for e in err:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
252 errors.append(e[0])
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
253 with open('inter.txt', 'w') as y:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
254 l = 0; a = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
255 for bb in bait:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
256 for lst in data:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
257 if proteins[a] not in errors:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
258 y.write(header[bait_index[l]] + '\t' + bb[1] + '\t' + proteins[a] + '\t'
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
259 + lst[bait_index[l]] + '\n')
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
260 a += 1
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
261 if a == len(proteins):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
262 l += 1; a = 0
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
263
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
264
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
265 def bait_check(bait, Scaffold_input):
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
266 # Check that bait names share Scaffold header titles.
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
267 bait_in = readtab(bait)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
268 header = read_Scaffold(Scaffold_input).header
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
269 for i in bait_in:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
270 if i[0] not in header:
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
271 sys.exit("Bait must share header titles with Scaffold output")
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
272
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
273 if __name__ == '__main__':
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
274 main(infile, baitfile)
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
275
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
276 os.rename("inter.txt", sys.argv[4])
019e60bd3f7f Uploaded
bornea
parents:
diff changeset
277 os.rename("bait.txt", sys.argv[7])