Mercurial > repos > rnateam > graphclust_postprocessing

annotate addCdhitseqs.py @ 17:f93c868203cc draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/GraphClust/CollectResults commit 4406735e44aba20859c252be39f4e99df28c7a92
author rnateam
date Sat, 27 Oct 2018 13:23:06 -0400
parents 869a6e807d76
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
1 import re
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
2 import glob
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
3 import sys
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
4
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
5 cdhitcluster = sys.argv[1]
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
6
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
7 cluster_seqs_stats_path = "RESULTS/*.cluster.all"
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
8 cluster_seqs_stats_files = glob.glob(cluster_seqs_stats_path)
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
9
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
10 repSeqRedSeqdict = {}
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
11 repLine = ""
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
12 count = 0
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
13 first = False
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
14 add_FullId = ""
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
15 k = 0
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
16
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
17 with open(cdhitcluster, 'r+') as f:
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
18 content = f.read()
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
19 reps = re.compile("^.*\*$", re.MULTILINE).findall(content)
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
20 lines = content.split('\n')
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
21
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
22 for i in range(0, len(lines)):
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
23 line = lines[i]
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
24 if ">Cluster" in line:
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
25 first = True
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
26 count = 0
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
27 repLine = reps[k]
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
28 k = k+1
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
29 continue
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
30 elif not first:
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
31 count += 1
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
32 first = False
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
33 else:
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
34 first = False
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
35 lineArr = []
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
36 if count > 0:
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
37 repLine = repLine.strip()
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
38 rep_FullId = repLine.split()[2]
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
39 rep_FullId = rep_FullId.replace(">","")
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
40 rep_FullId = rep_FullId.replace("...","")
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
41 if "*" in line or not line.strip():
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
42 continue
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
43 line = line.strip()
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
44 add_FullId = line.split()[2]
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
45 add_FullId = add_FullId.replace(">","")
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
46 add_FullId = add_FullId.replace("...","")
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
47 lineArr.append(add_FullId)
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
48 repSeqRedSeqdict[rep_FullId] = lineArr
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
49
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
50 toWrite = ""
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
51 for singleFile in sorted(cluster_seqs_stats_files):
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
52 toWrite = ""
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
53 with open(singleFile, "r+") as clFile:
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
54 file_lines = clFile.readlines()
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
55 for line in file_lines:
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
56 line = '\t'.join(line.split())
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
57 toWrite += line + '\n'
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
58 clFile.seek(0)
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
59 clFile.write(toWrite)
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
60 clFile.truncate()
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
61 first_line = file_lines[0]
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
62 toWrite = ""
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
63 cols = first_line.split()
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
64 file_content = '\n'.join(file_lines)
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
65 for key, val in repSeqRedSeqdict.items():
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
66 if key in file_content:
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
67
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
68 for i in val:
6
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
69 cols[3] = "---"
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
70 cols[4] = "CD-Hit"
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
71 cols[7] = str(i)
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
72 if len(first_line.split()) > 9:
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
73 cols[9] = str(i.rsplit("_",1)[0])
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
74 toWrite += '\t'.join(cols)
869a6e807d76 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit 057c2fd398055dc86eb2c00d8a74f301d5c231d9-dirty
rnateam
parents: 3
diff changeset
75 toWrite +="\n"
3
79b9117aef01 planemo upload for repository https://github.com/eteriSokhoyan/galaxytools/tree/branchForIterations/tools/GraphClust/CollectResults commit c03cf64554289eb098267c0923cf0cf7b245cc0c
rnateam
parents:
diff changeset
76 clFile.write(toWrite)