Mercurial > repos > jay > pdaug_addclasslabel
annotate PDAUG_TSVtoFASTA/PDAUG_TSVtoFASTA.py @ 1:8c2ee62ce9d8 draft
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
| author | jay | 
|---|---|
| date | Tue, 29 Dec 2020 04:48:30 +0000 | 
| parents | 2df11ea23f10 | 
| children | 3ef2416c21c3 | 
| rev | line source | 
|---|---|
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
1 import os | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
2 import argparse | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
3 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
4 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
5 def TSVtoFASTA(InFile, Method, Positive, Negative, OutFile): | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
6 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
7 if Method == 'WithClassLabel': | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
8 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
9 f = open(InFile) | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
10 lines = f.readlines() | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
11 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
12 of1 = open(Positive,'w') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
13 of2 = open(Negative,'w') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
14 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
15 n = 0 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
16 m = 0 | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
17 | 
| 
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
18 l = [] | 
| 
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
19 | 
| 
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
20 for line in lines[1:]: | 
| 
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
21 l.append(line.split('\t')[1].strip('\n').strip('\r')) | 
| 
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
22 l = list(set(l)) | 
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
23 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
24 for line in lines: | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
25 | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
26 if l[0] in line.split('\t')[1].strip('\n').strip('\r'): | 
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
27 n= n+1 | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
28 of1.write('>peptide_'+str(n)+'_'+str(l[0])+'\n') | 
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
29 of1.write(line.split('\t')[0]+'\n') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
30 | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
31 if l[1] in line.split('\t')[1].strip('\n').strip('\r'): | 
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
32 m= m+1 | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
33 of2.write('>peptide_'+str(m)+'_'+str(l[1])+'\n') | 
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
34 of2.write(line.split('\t')[0]+'\n') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
35 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
36 elif Method == 'NoClassLabel': | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
37 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
38 f = open(InFile) | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
39 lines = f.readlines() | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
40 of1 = open(OutFile,'w') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
41 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
42 for i, line in enumerate(lines[1:]): | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
43 of1.write('>peptide_'+str(i)+'\n') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
44 of1.write(line.split('\t')[0]+'\n') | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
45 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
46 else: | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
47 pass | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
48 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
49 if __name__=="__main__": | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
50 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
51 import argparse | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
52 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
53 parser = argparse.ArgumentParser() | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
54 | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
55 parser.add_argument("-I", "--InFile", required=True, default=None, help=".fasta or .tsv") | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
56 parser.add_argument("-P", "--Postvs", required=False, default='FirstDataFile.fasta', help="Path to target tsv file") | 
| 
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
57 parser.add_argument("-N", "--Negtvs", required=False, default='SecondDataFile.fasta', help="Path to target tsv file") | 
| 
0
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
58 parser.add_argument("-O", "--OutFile", required=False, default='OutFile.fasta', help="Path to target tsv file") | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
59 parser.add_argument("-M", "--Method", required=True, default=None, help="Path to target tsv file") | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
60 args = parser.parse_args() | 
| 
 
2df11ea23f10
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
 
jay 
parents:  
diff
changeset
 | 
61 | 
| 
1
 
8c2ee62ce9d8
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 3c91f421d26c8f42cf2671e47db735d2cf69dde8"
 
jay 
parents: 
0 
diff
changeset
 | 
62 TSVtoFASTA(args.InFile, args.Method, args.Postvs, args.Negtvs, args.OutFile) | 
