Mercurial > repos > jay > pdaug_merge_dataframes

annotate PDAUG_TSVtoFASTA/PDAUG_TSVtoFASTA.py @ 8:4056124db7c8 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit d396d7ff89705cc0dd626ed32c45a9f4029b1b05"
author jay
date Mon, 10 Jan 2022 04:23:30 +0000
parents b5a9f541c6f8
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
1
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
2 import pandas as pd
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
3
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
5 def TSVtoFASTA(infile, method, firstdatafile, seconddatafile, outfile, clmpepid, slcclasslabel, peps):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
6
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
7
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
8 fn = [firstdatafile, seconddatafile]
0
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
9
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
10
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
11 df = pd.read_csv(infile, sep="\t")
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
12 if clmpepid == None:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
13 pass
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
14 else:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
15 names = df[clmpepid].tolist()
0
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
16
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
17 peps = df[peps].tolist()
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
18
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
19 if method == "withoutlabel":
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
20 f = open(outfile,'w')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
21 if clmpepid is not None:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
22 for i,n in enumerate(peps):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
23 f.write(">"+names[i]+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
24 f.write(n+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
25 f.close()
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
26 else:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
27 for i,n in enumerate(peps):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
28 f.write(">"+str(i)+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
29 f.write(n+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
30 f.close()
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
31
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
32 elif method == "withlabel":
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
33 labels = df[slcclasslabel].tolist()
0
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
34
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
35 label = list(set(labels))
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
36
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
37 if clmpepid is None:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
38 for i, l in enumerate(label):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
39 f = open(fn[i],'w')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
40 print('ok1')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
41 for i, L in enumerate(labels):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
42 if l == L:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
43 f.write(">"+str(i)+"_"+str(l)+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
44 f.write(peps[i]+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
45 f.close()
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
46 else:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
47 for i, l in enumerate(label):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
48 f = open(fn[i],'w')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
49 for i, L in enumerate(labels):
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
50 if l == L:
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
51 f.write(">"+names[i]+"_"+l+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
52 f.write(peps[i]+'\n')
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
53 f.close()
0
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
54
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
55 if __name__=="__main__":
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
56
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
57 import argparse
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
58 parser = argparse.ArgumentParser()
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
59 parser.add_argument("-I", "--InFile", required=True, default=None, help=".fasta or .tsv")
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
60 parser.add_argument("-F", "--FirstDataFile", required=False, default='FirstDataFile.fasta', help="Path to target tsv file")
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
61 parser.add_argument("-S", "--SecondDataFile", required=False, default='SecondDataFile.fasta', help="Path to target tsv file")
0
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
62 parser.add_argument("-O", "--OutFile", required=False, default='OutFile.fasta', help="Path to target tsv file")
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
63 parser.add_argument("-M", "--Method", required=True, default=None, help="Path to target tsv file")
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
64 parser.add_argument("-C", "--ClmPepID", required=False, default=None, help="Peptide Column Name")
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
65 parser.add_argument("-L", "--SlcClassLabel", required=False, default="Class_label", help="Class Label Column Name")
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
66 parser.add_argument("-P", "--PeptideColumn", required=True, default=None, help="Class Label Column Name")
0
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
67 args = parser.parse_args()
5bb52d4bf172 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit a9bd83f6a1afa6338cb6e4358b63ebff5bed155e"
jay
parents:
diff changeset
68
4
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
69 TSVtoFASTA(args.InFile, args.Method, args.FirstDataFile, args.SecondDataFile, args.OutFile, args.ClmPepID, args.SlcClassLabel, args.PeptideColumn)
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
70
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
71
b5a9f541c6f8 "planemo upload for repository https://github.com/jaidevjoshi83/pdaug commit 45ebf32dcaa1eed91670d3a2491f9cf3dfb535ef"
jay
parents: 3
diff changeset
72