Mercurial > repos > rmarenco > hubarchivecreator
annotate Gtf.py @ 25:99dad5f9444c draft
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 93e2e2fb59f99677425104a80c17f665fa7b2b4a-dirty
author | yating-l |
---|---|
date | Fri, 02 Jun 2017 17:36:24 -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 util import subtools |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
9 |
19 | 10 class InfoModifiedGtf(): |
11 def __init__(self, is_modified=False, array_modified_lines=[]): | |
12 self.is_modified = is_modified | |
13 self.array_modified_lines = array_modified_lines | |
14 | |
15 def get_str_modified_lines(self): | |
16 return ','.join(map(str, self.array_modified_lines)) | |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
17 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
18 class Gtf( Datatype ): |
10
acc233161f50
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents:
1
diff
changeset
|
19 def __init__( self, input_gtf_false_path, data_gtf): |
acc233161f50
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents:
1
diff
changeset
|
20 |
acc233161f50
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1b1063f90004764bcf504f4340738eca5c4b1f9d
rmarenco
parents:
1
diff
changeset
|
21 super(Gtf, self).__init__() |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
22 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
23 self.track = None |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
24 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
25 self.input_gtf_false_path = input_gtf_false_path |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
26 self.name_gtf = data_gtf["name"] |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
27 self.priority = data_gtf["order_index"] |
16
3233451a3bd6
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit fc73ec22a0db3ab09c4ac13dc58f0b54ae37845c
rmarenco
parents:
13
diff
changeset
|
28 self.track_color = data_gtf["track_color"] |
17
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
29 # 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
|
30 self.group_name = data_gtf["group_name"] |
23
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
19
diff
changeset
|
31 if data_gtf["long_label"]: |
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
19
diff
changeset
|
32 self.long_label = data_gtf["long_label"] |
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
19
diff
changeset
|
33 else: |
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
19
diff
changeset
|
34 self.long_label = self.name_gtf |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
35 #print "Creating TrackHub GTF from (falsePath: %s; name: %s)" % ( self.input_gtf_false_path, self.name_gtf) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
36 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
37 # TODO: See if we need these temporary files as part of the generated files |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
38 genePredFile = tempfile.NamedTemporaryFile(bufsize=0, suffix=".genePred") |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
39 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
|
40 sorted_bigGenePred_file = tempfile.NamedTemporaryFile(suffix=".sortedBed.bigGenePred") |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
41 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
42 # GtfToGenePred |
19 | 43 ## Checking the integrity of the inputs |
44 modified_gtf = self._checkAndFixGtf() | |
45 | |
46 ## Processing the gtf | |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
47 subtools.gtfToGenePred(self.input_gtf_false_path, genePredFile.name) |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
48 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
49 # TODO: From there, refactor because common use with Gff3.py |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
50 # genePredToBigGenePred processing |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
51 subtools.genePredToBigGenePred(genePredFile.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
|
52 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
53 # Sort processing |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
54 subtools.sort(unsorted_bigGenePred_file.name, sorted_bigGenePred_file.name) |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
55 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
56 # bedToBigBed processing |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
57 trackName = "".join( ( self.name_gtf, ".bb") ) |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
58 |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
59 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
|
60 |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
61 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
|
62 |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
63 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
|
64 subtools.bedToBigBed(sorted_bigGenePred_file.name, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
65 self.chromSizesFile.name, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
66 bigBedFile.name, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
67 autoSql=auto_sql_option, |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
68 typeOption='bed12+8', |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
69 tab=True) |
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
70 |
1
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
71 |
fb5e60d4d18a
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 64cfc08088d11f6818c1b4e5514ef9e67969eaff-dirty
rmarenco
parents:
diff
changeset
|
72 # Create the Track Object |
11
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
73 self.createTrack(file_path=trackName, |
d05236b15f81
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 3760d0c8353b924ecf994131a5c2eb381aa81fb2
rmarenco
parents:
10
diff
changeset
|
74 track_name=trackName, |
23
2677f1899aa8
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 07210e24fe7d22f5dfc84061310280f67f781fb1-dirty
yating-l
parents:
19
diff
changeset
|
75 long_label=self.long_label, track_type='bigGenePred', |
13
25809f699cb3
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 65ab931ef2b05a5acf06cbde3a746c94a0a0a4cb
rmarenco
parents:
11
diff
changeset
|
76 visibility='dense', priority=self.priority, |
16
3233451a3bd6
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit fc73ec22a0db3ab09c4ac13dc58f0b54ae37845c
rmarenco
parents:
13
diff
changeset
|
77 track_file=myBigBedFilePath, |
17
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
78 track_color=self.track_color, |
c02720d1afee
planemo upload for repository https://github.com/goeckslab/hub-archive-creator commit 1adbf397de1fc7af4d91e026093d7fff983e21cf
rmarenco
parents:
16
diff
changeset
|
79 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
|
80 |
19 | 81 # TODO: Use Logging instead of print |
82 if modified_gtf.is_modified: | |
83 print("- Warning: Gtf %s created with a modified version of your Gtf because of start/end coordinates issues." | |
84 % self.name_gtf) | |
85 print("Here are the lines removed: " + modified_gtf.get_str_modified_lines()) | |
86 else: | |
87 print("- Gtf %s created" % self.name_gtf) | |
88 | |
89 def _checkAndFixGtf(self): | |
90 """ | |
91 Call _checkAndFixGtf, check the integrity of gtf file, | |
92 if coordinates exceed chromosome size, either removed the whole line(s) or truncated to the end of the scaffold | |
93 depending on the user choice | |
94 default: remove the whole line(s) | |
95 """ | |
96 # Set the boolean telling if we had to modify the file | |
97 modified_gtf = InfoModifiedGtf() | |
98 | |
99 # Create a temp gtf just in case we have issues | |
100 temp_gtf = tempfile.NamedTemporaryFile(bufsize=0, suffix=".gtf", delete=False) | |
101 | |
102 # TODO: Get the user choice and use it | |
103 # TODO: Check if the start > 0 and the end <= chromosome size | |
104 # Get the chrom.sizes into a dictionary to have a faster access | |
105 # TODO: Think about doing this in Datatype.py, so everywhere we have access to this read-only dictionary | |
106 dict_chrom_sizes = {} | |
107 with open(self.chromSizesFile.name, 'r') as chromSizes: | |
108 lines = chromSizes.readlines() | |
109 for line in lines: | |
110 fields = line.split() | |
111 # fields[1] should be the name of the scaffold | |
112 # fields[2] should be the size of the scaffold | |
113 # TODO: Ensure this is true for all lines | |
114 dict_chrom_sizes[fields[0]] = fields[1] | |
115 | |
116 # Parse the GTF and check each line using the chrom sizes dictionary | |
117 with open(temp_gtf.name, 'a+') as tmp: | |
118 with open(self.input_gtf_false_path, 'r') as gtf: | |
119 lines = gtf.readlines() | |
120 for index, line in enumerate(lines): | |
121 # If this is not a comment, we check the fields | |
122 if not line.startswith('#'): | |
123 fields = line.split() | |
124 # We are interested in fields[0] => Seqname (scaffold) | |
125 # We are interested in fields[3] => Start of the scaffold | |
126 # We are interested in fields[4] => End of the scaffold | |
127 scaffold_size = dict_chrom_sizes[fields[0]] | |
128 start_position = fields[3] | |
129 end_position = fields[4] | |
130 | |
131 if start_position > 0 and end_position <= scaffold_size: | |
132 # We are good, so we copy this line | |
133 tmp.write(line) | |
134 tmp.write(os.linesep) | |
135 | |
136 | |
137 # The sequence is not good, we are going to process it regarding the user choice | |
138 # TODO: Process the user choice | |
139 # By default, we are assuming the user choice is to remove the lines: We don't copy it | |
140 | |
141 # If we are here, it means the gtf has been modified | |
142 else: | |
143 # We save the line for the feedback to the user | |
144 modified_gtf.array_modified_lines.append(index + 1) | |
145 | |
146 if modified_gtf.is_modified is False: | |
147 modified_gtf.is_modified = True | |
148 else: | |
149 pass | |
150 else: | |
151 tmp.write(line) | |
152 tmp.write(os.linesep) | |
153 | |
154 # Once the process it completed, we just replace the path of the gtf | |
155 self.input_gtf_false_path = temp_gtf.name | |
156 | |
157 # TODO: Manage the issue with the fact the dataset is going to still exist on the disk because of delete=False | |
158 | |
159 return modified_gtf |