mirror of
https://github.com/galaxyproject/galaxy.git
synced 2026-10-16 15:45:04 +08:00
Daniel Blankenberg
87eebd3206
These changes should only affect output when an input block has a species appearing more than once, with the exception being the MAF to multiple FASTA blocks converters: the FASTA headers have been revised to included the sequence index for a species in a block as well as the block index. A new tool "Split MAF Blocks by Species" has been added that will split MAF blocks into the complete combination of multiple blocks when a species appears more than once.
45 lines
1.4 KiB
Python
45 lines
1.4 KiB
Python
#!/usr/bin/env python
|
|
|
|
"""
|
|
Read a maf and split blocks by unique species combinations
|
|
"""
|
|
import sys
|
|
from galaxy import eggs
|
|
import pkg_resources; pkg_resources.require( "bx-python" )
|
|
from bx.align import maf
|
|
from galaxy.tools.util import maf_utilities
|
|
from galaxy.util import string_as_bool
|
|
|
|
assert sys.version_info[:2] >= ( 2, 4 )
|
|
|
|
def __main__():
|
|
try:
|
|
maf_reader = maf.Reader( open( sys.argv[1] ) )
|
|
except Exception, e:
|
|
maf_utilities.tool_fail( "Error opening MAF: %s" % e )
|
|
try:
|
|
out = maf.Writer( open( sys.argv[2], "w") )
|
|
except Exception, e:
|
|
maf_utilities.tool_fail( "Error opening file for output: %s" % e )
|
|
try:
|
|
collapse_columns = string_as_bool( sys.argv[3] )
|
|
except Exception, e:
|
|
maf_utilities.tool_fail( "Error determining collapse columns value: %s" % e )
|
|
|
|
start_count = 0
|
|
end_count = 0
|
|
for start_count, start_block in enumerate( maf_reader ):
|
|
for block in maf_utilities.iter_blocks_split_by_species( start_block ):
|
|
if collapse_columns:
|
|
block.remove_all_gap_columns()
|
|
out.write( block )
|
|
end_count += 1
|
|
out.close()
|
|
|
|
if end_count:
|
|
print "%i alignment blocks created from %i original blocks." % ( end_count, start_count + 1 )
|
|
else:
|
|
print "No alignment blocks were created."
|
|
|
|
if __name__ == "__main__": __main__()
|