Mercurial > repos > devteam > subtract_query
changeset 1:df4bce472400 draft
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
| author | devteam |
|---|---|
| date | Tue, 13 Oct 2015 12:51:35 -0400 |
| parents | 8a3448727cec |
| children | 659a1c095357 |
| files | subtract_query.py tool_dependencies.xml |
| diffstat | 2 files changed, 21 insertions(+), 17 deletions(-) [+] |
line wrap: on
line diff
--- a/subtract_query.py Tue Apr 01 09:12:49 2014 -0400 +++ b/subtract_query.py Tue Oct 13 12:51:35 2015 -0400 @@ -6,7 +6,7 @@ usage: %prog in_file_1 in_file_2 begin_col end_col output --ignore-empty-end-cols: ignore empty end columns when subtracting """ -import sys, re +import sys from bx.cookbook import doc_optparse # Older py compatibility @@ -17,6 +17,7 @@ assert sys.version_info[:2] >= ( 2, 4 ) + def get_lines(fname, begin_col='', end_col='', ignore_empty_end_cols=False): lines = set([]) i = 0 @@ -27,22 +28,25 @@ """Both begin_col and end_col must be integers at this point.""" try: line = line.split('\t') - line = '\t'.join([line[j] for j in range(begin_col-1, end_col)]) + line = '\t'.join([line[j] for j in range(begin_col - 1, end_col)]) if ignore_empty_end_cols: # removing empty fields, we do not compare empty fields at the end of a line. line = line.rstrip() lines.add( line ) - except: pass + except: + pass else: if ignore_empty_end_cols: # removing empty fields, we do not compare empty fields at the end of a line. line = line.rstrip() lines.add( line ) - if i: return (i+1, lines) - else: return (i, lines) + if i: + return (i + 1, lines) + else: + return (i, lines) + def main(): - # Parsing Command Line here options, args = doc_optparse.parse( __doc__ ) @@ -50,10 +54,10 @@ inp1_file, inp2_file, begin_col, end_col, out_file = args except: doc_optparse.exception() - + begin_col = begin_col.strip() end_col = end_col.strip() - + if begin_col != 'None' or end_col != 'None': """ The user selected columns for restriction. We'll allow default @@ -75,7 +79,7 @@ begin_col = end_col = '' try: - fo = open(out_file,'w') + fo = open(out_file, 'w') except: print >> sys.stderr, "Unable to open output file" sys.exit() @@ -88,7 +92,7 @@ len1, lines1 = get_lines(inp1_file, begin_col, end_col, options.ignore_empty_end_cols) diff1 = len1 - len(lines1) len2, lines2 = get_lines(inp2_file, begin_col, end_col, options.ignore_empty_end_cols) - + lines1.difference_update(lines2) """lines1 is now the set of unique lines in inp1_file - the set of unique lines in inp2_file""" @@ -96,15 +100,15 @@ print >> fo, line fo.close() - - info_msg = 'Subtracted %d lines. ' %((len1 - diff1) - len(lines1)) - + + info_msg = 'Subtracted %d lines. ' % ((len1 - diff1) - len(lines1)) + if begin_col and end_col: info_msg += 'Restricted to columns c' + str(begin_col) + ' thru c' + str(end_col) + '. ' if diff1 > 0: - info_msg += 'Eliminated %d duplicate/blank/comment/invalid lines from first query.' %diff1 - + info_msg += 'Eliminated %d duplicate/blank/comment/invalid lines from first query.' % diff1 + print info_msg if __name__ == "__main__":
--- a/tool_dependencies.xml Tue Apr 01 09:12:49 2014 -0400 +++ b/tool_dependencies.xml Tue Oct 13 12:51:35 2015 -0400 @@ -1,9 +1,9 @@ <?xml version="1.0"?> <tool_dependency> <package name="bx-python" version="0.7.1"> - <repository changeset_revision="cdb5991f9790" name="package_bx_python_0_7" owner="devteam" prior_installation_required="False" toolshed="http://testtoolshed.g2.bx.psu.edu" /> + <repository changeset_revision="35e2457234ef" name="package_bx_python_0_7" owner="devteam" toolshed="https://testtoolshed.g2.bx.psu.edu" /> </package> <package name="galaxy-ops" version="1.0.0"> - <repository changeset_revision="3287c55c02b8" name="package_galaxy_ops_1_0_0" owner="devteam" prior_installation_required="False" toolshed="http://testtoolshed.g2.bx.psu.edu" /> + <repository changeset_revision="3287c55c02b8" name="package_galaxy_ops_1_0_0" owner="devteam" toolshed="https://testtoolshed.g2.bx.psu.edu" /> </package> </tool_dependency>