Mercurial > repos > matt-shirley > sra_tools

annotate sra.py @ 13:45031bbf6b27 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
better sra_fetch code, compliant (?) datatype, updated dependencies in tool wrappers
author Matt Shirley <mdshw5@gmail.com>
date Mon, 17 Jun 2013 16:17:33 -0400
parents b77840618b8f
children 93a60318b9ca
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
1 """
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
2 Sra class
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
3 """
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
4
12
b77840618b8f debugging SRA datatype
Matt Shirley <mdshw5@gmail.com>
parents: 1
diff changeset
5 import binascii
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
6 from galaxy.datatypes.binary import Binary
12
b77840618b8f debugging SRA datatype
Matt Shirley <mdshw5@gmail.com>
parents: 1
diff changeset
7 from galaxy.datatypes.data import Data
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
8 from galaxy.datatypes.metadata import MetadataElement
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
9 from galaxy.datatypes.sniff import *
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
10
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
11 class Sra( Binary ):
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
12 """ Sequence Read Archive (SRA) """
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
13 file_ext = 'sra'
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
14
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
15 def __init__( self, **kwd ):
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
16 Binary.__init__( self, **kwd )
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
17 def sniff( self, filename ):
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
18 """ The first 8 bytes of any NCBI sra file is 'NCIB.sra', and the file is binary. Not sure if EBI and DDBJ files may differ.
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
19 For details about the format, see http://www.ncbi.nlm.nih.gov/books/n/helpsra/SRA_Overview_BK/#SRA_Overview_BK.4_SRA_Data_Structure
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
20 """
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
21 try:
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
22 header = open(filename).read(8)
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
23 if binascii.b2a_hex(header) == binascii.hexlify('NCBI.sra'):
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
24 return True
12
b77840618b8f debugging SRA datatype
Matt Shirley <mdshw5@gmail.com>
parents: 1
diff changeset
25 else:
b77840618b8f debugging SRA datatype
Matt Shirley <mdshw5@gmail.com>
parents: 1
diff changeset
26 return False
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
27 except:
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
28 return False
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
29 def set_peek(self, dataset, is_multi_byte=False):
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
30 if not dataset.dataset.purged:
12
b77840618b8f debugging SRA datatype
Matt Shirley <mdshw5@gmail.com>
parents: 1
diff changeset
31 dataset.peek = 'Binary SRA file'
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
32 dataset.blurb = data.nice_size(dataset.get_size())
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
33 else:
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
34 dataset.peek = 'file does not exist'
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
35 dataset.blurb = 'file purged from disk'
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
36 def display_peek(self, dataset):
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
37 try:
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
38 return dataset.peek
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
39 except:
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
40 return 'Binary sra file (%s)' % ( data.nice_size(dataset.get_size()))