Mercurial > repos > rmarenco > hubarchivecreator
annotate Gff3.py @ 24:fcc1021bd496 draft
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 93e2e2fb59f99677425104a80c17f665fa7b2b4a-dirty
author | yating-l |
---|---|
date | Wed, 31 May 2017 11:35:16 -0400 |
parents | 2677f1899aa8 |
children | df42241d3731 |
rev | line source |
---|---|
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
1 #!/usr/bin/python |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
2 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
3 import os |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
4 import tempfile |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
5 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
6 # Internal dependencies |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
7 from Datatype import Datatype |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
8 from Track import Track |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
9 from TrackDb import TrackDb |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
10 from util import subtools |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
11 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
12 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
13 class Gff3( Datatype ): |
10
acc233161f50
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents:
1
diff
changeset
|
14 def __init__(self, input_Gff3_false_path, data_gff3): |
acc233161f50
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents:
1
diff
changeset
|
15 super( Gff3, self ).__init__() |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
16 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
17 self.track = None |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
18 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
19 self.input_Gff3_false_path = input_Gff3_false_path |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
20 self.name_gff3 = data_gff3["name"] |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
21 self.priority = data_gff3["order_index"] |
16
3233451a3bd6
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit fc73ec22a0db3ab09c4ac13dc58f0b54ae37845c
rmarenco
parents:
13
diff
changeset
|
22 self.track_color = data_gff3["track_color"] |
17
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
23 # TODO: Think about how to avoid repetition of the group_name everywhere |
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
24 self.group_name = data_gff3["group_name"] |
23
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
17
diff
changeset
|
25 if data_gff3["long_label"]: |
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
17
diff
changeset
|
26 self.long_label = data_gff3["long_label"] |
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
17
diff
changeset
|
27 else: |
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
17
diff
changeset
|
28 self.long_label = self.name_gff3 |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
29 # TODO: See if we need these temporary files as part of the generated files |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
30 unsorted_genePred_file = tempfile.NamedTemporaryFile(bufsize=0, suffix=".genePred") |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
31 unsorted_bigGenePred_file = tempfile.NamedTemporaryFile(bufsize=0, suffix=".unsorted.bigGenePred") |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
32 sorted_biGenePred_file = tempfile.NamedTemporaryFile(suffix=".sorted.bigGenePred") |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
33 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
34 # gff3ToGenePred processing |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
35 subtools.gff3ToGenePred(self.input_Gff3_false_path, unsorted_genePred_file.name) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
36 |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
37 # genePredToBigGenePred |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
38 subtools.genePredToBigGenePred(unsorted_genePred_file.name, unsorted_bigGenePred_file.name) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
39 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
40 # Sort processing |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
41 subtools.sort(unsorted_bigGenePred_file.name, sorted_biGenePred_file.name) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
42 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
43 # TODO: Check if no errors |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
44 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
45 # bedToBigBed processing |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
46 trackName = "".join( (self.name_gff3, ".bb" ) ) |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
47 |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
48 auto_sql_option = os.path.join(self.tool_directory, 'bigGenePred.as') |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
49 |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
50 myBigBedFilePath = os.path.join(self.myTrackFolderPath, trackName) |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
51 |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
52 with open(myBigBedFilePath, 'w') as bigBedFile: |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
53 subtools.bedToBigBed(sorted_biGenePred_file.name, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
54 self.chromSizesFile.name, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
55 bigBedFile.name, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
56 autoSql=auto_sql_option, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
57 typeOption='bed12+8', |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
58 tab=True) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
59 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
60 # Create the Track Object |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
61 self.createTrack(file_path=trackName, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
62 track_name=trackName, |
23
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
17
diff
changeset
|
63 long_label=self.long_label, |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
64 track_type='bigGenePred', visibility='dense', |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
65 priority=self.priority, |
16
3233451a3bd6
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit fc73ec22a0db3ab09c4ac13dc58f0b54ae37845c
rmarenco
parents:
13
diff
changeset
|
66 track_file=myBigBedFilePath, |
17
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
67 track_color=self.track_color, |
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
68 group_name=self.group_name) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
69 |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
70 print("- Gff3 %s created" % self.name_gff3) |