0
|
1 """
|
|
2 SnpEff datatypes
|
|
3 """
|
|
4 import os,os.path,re,sys
|
|
5 import galaxy.datatypes.data
|
|
6 from galaxy.datatypes.data import Text
|
|
7 from galaxy.datatypes.metadata import MetadataElement
|
|
8
|
|
9 class SnpEffDb( Text ):
|
|
10 """Class describing an IGV tiled data file (TDF) .tdf binary file"""
|
|
11 file_ext = "snpeffdb"
|
|
12 MetadataElement( name="genome_version", default=None, desc="Genome Version", readonly=True, visible=True, no_value=None )
|
|
13 MetadataElement( name="regulation", default=[], desc="Regulation Names", readonly=True, visible=True, no_value=[] )
|
|
14 MetadataElement( name="annotation", default=[], desc="Annotation Names", readonly=True, visible=True, no_value=[] )
|
|
15
|
|
16 def __init__( self, **kwd ):
|
|
17 Text.__init__( self, **kwd )
|
|
18
|
|
19 def set_meta( self, dataset, **kwd ):
|
|
20 Text.set_meta(self, dataset, **kwd )
|
|
21 data_dir = dataset.extra_files_path
|
|
22 ## search data_dir/genome_version for files
|
|
23 regulation_pattern = 'regulation_(.+).bin'
|
|
24 # annotation files that are included in snpEff by a flag
|
|
25 annotations_dict = {'nextProt.bin' : '-nextprot','motif.bin': '-motif'}
|
|
26 regulations = []
|
|
27 annotations = []
|
|
28 if data_dir and os.path.isdir(data_dir):
|
|
29 for root, dirs, files in os.walk(data_dir):
|
|
30 for fname in files:
|
|
31 if fname.startswith('snpEffectPredictor'):
|
|
32 # if snpEffectPredictor.bin download succeeded
|
|
33 genome_version = os.path.basename(root)
|
|
34 dataset.metadata.genome_version = genome_version
|
|
35 else:
|
|
36 m = re.match(regulation_pattern,fname)
|
|
37 if m:
|
|
38 name = m.groups()[0]
|
|
39 regulations.append(name)
|
|
40 elif fname in annotations_dict:
|
|
41 value = annotations_dict[fname]
|
|
42 name = value.lstrip('-')
|
|
43 annotations.append(name)
|
|
44 dataset.metadata.regulation = regulations
|
|
45 dataset.metadata.annotation = annotations
|
|
46
|