| Previous changeset 1:8ff0ac66f1a3 (2015-05-13) Next changeset 3:481b0a925e66 (2017-05-17) |
|
Commit message:
v0.0.5 Depend on Biopython 1.67 from Tool Shed or (Bio)conda |
|
modified:
tools/seq_filter_by_mapping/README.rst tools/seq_filter_by_mapping/seq_filter_by_mapping.py tools/seq_filter_by_mapping/seq_filter_by_mapping.xml tools/seq_filter_by_mapping/tool_dependencies.xml |
| b |
| diff -r 8ff0ac66f1a3 -r 48e71dfd51b3 tools/seq_filter_by_mapping/README.rst --- a/tools/seq_filter_by_mapping/README.rst Wed May 13 11:08:58 2015 -0400 +++ b/tools/seq_filter_by_mapping/README.rst Wed May 10 13:16:44 2017 -0400 |
| b |
| @@ -1,7 +1,7 @@ Galaxy tool to filter FASTA, FASTQ or SFF sequences by SAM/BAM mapping ====================================================================== -This tool is copyright 2014-2015 by Peter Cock, The James Hutton Institute +This tool is copyright 2014-2017 by Peter Cock, The James Hutton Institute (formerly SCRI, Scottish Crop Research Institute), UK. All rights reserved. See the licence text below. @@ -66,6 +66,10 @@ v0.0.4 - Use the ``format_source=...`` tag. - Reorder XML elements (internal change only). - Planemo for Tool Shed upload (``.shed.yml``, internal change only). +v0.0.5 - Python script cleanups (internal change only). + - Depends on Biopython 1.67 via legacy Tool Shed package or bioconda. + - Use ``<command detect_errors="aggressive">`` (internal change only). + - Single quote command line arguments (internal change only). ======= ====================================================================== @@ -82,17 +86,17 @@ Planemo commands (which requires you have set your Tool Shed access details in ``~/.planemo.yml`` and that you have access rights on the Tool Shed):: - $ planemo shed_upload --shed_target testtoolshed --check_diff ~/repositories/pico_galaxy/tools/seq_filter_by_mapping/ + $ planemo shed_update -t testtoolshed --check_diff tools/seq_filter_by_mapping/ ... or:: - $ planemo shed_upload --shed_target toolshed --check_diff ~/repositories/pico_galaxy/tools/seq_filter_by_mapping/ + $ planemo shed_update -t toolshed --check_diff tools/seq_filter_by_mapping/ ... To just build and check the tar ball, use:: - $ planemo shed_upload --tar_only ~/repositories/pico_galaxy/tools/seq_filter_by_mapping/ + $ planemo shed_upload --tar_only tools/seq_filter_by_mapping/ ... $ tar -tzf shed_upload.tar.gz test-data/SRR639755_mito_pairs.fastq.gz |
| b |
| diff -r 8ff0ac66f1a3 -r 48e71dfd51b3 tools/seq_filter_by_mapping/seq_filter_by_mapping.py --- a/tools/seq_filter_by_mapping/seq_filter_by_mapping.py Wed May 13 11:08:58 2015 -0400 +++ b/tools/seq_filter_by_mapping/seq_filter_by_mapping.py Wed May 10 13:16:44 2017 -0400 |
| [ |
| b'@@ -18,17 +18,15 @@\n \n Use -v or --version to get the version, -h or --help for help.\n """\n+\n import os\n-import sys\n import re\n import subprocess\n+import sys\n+\n from optparse import OptionParser\n \n-def sys_exit(msg, err=1):\n- sys.stderr.write(msg.rstrip() + "\\n")\n- sys.exit(err)\n-\n-#Parse Command Line\n+# Parse Command Line\n usage = """Use as follows:\n \n $ python seq_filter_by_mapping.py [options] mapping.sam/bam [more mappings]\n@@ -64,7 +62,7 @@\n options, args = parser.parse_args()\n \n if options.version:\n- print "v0.0.3"\n+ print "v0.0.5"\n sys.exit(0)\n \n in_file = options.input\n@@ -74,27 +72,27 @@\n pair_mode = options.pair_mode\n \n if in_file is None or not os.path.isfile(in_file):\n- sys_exit("Missing input file: %r" % in_file)\n+ sys.exit("Missing input file: %r" % in_file)\n if out_positive_file is None and out_negative_file is None:\n- sys_exit("Neither output file requested")\n+ sys.exit("Neither output file requested")\n if seq_format is None:\n- sys_exit("Missing sequence format")\n+ sys.exit("Missing sequence format")\n if pair_mode not in ["lax", "strict"]:\n- sys_exit("Pair mode argument should be \'lax\' or \'strict\', not %r" % pair_mode)\n+ sys.exit("Pair mode argument should be \'lax\' or \'strict\', not %r" % pair_mode)\n for mapping in args:\n if not os.path.isfile(mapping):\n- sys_exit("Mapping file %r not found" % mapping)\n+ sys.exit("Mapping file %r not found" % mapping)\n if not args:\n- sys_exit("At least one SAM/BAM mapping file is required")\n+ sys.exit("At least one SAM/BAM mapping file is required")\n \n \n-#Cope with three widely used suffix naming convensions,\n-#Illumina: /1 or /2\n-#Forward/revered: .f or .r\n-#Sanger, e.g. .p1k and .q1k\n-#See http://staden.sourceforge.net/manual/pregap4_unix_50.html\n-#re_f = re.compile(r"(/1|\\.f|\\.[sfp]\\d\\w*)$")\n-#re_r = re.compile(r"(/2|\\.r|\\.[rq]\\d\\w*)$")\n+# Cope with three widely used suffix naming convensions,\n+# Illumina: /1 or /2\n+# Forward/revered: .f or .r\n+# Sanger, e.g. .p1k and .q1k\n+# See http://staden.sourceforge.net/manual/pregap4_unix_50.html\n+# re_f = re.compile(r"(/1|\\.f|\\.[sfp]\\d\\w*)$")\n+# re_r = re.compile(r"(/2|\\.r|\\.[rq]\\d\\w*)$")\n re_suffix = re.compile(r"(/1|\\.f|\\.[sfp]\\d\\w*|/2|\\.r|\\.[rq]\\d\\w*)$")\n assert re_suffix.search("demo.f")\n assert re_suffix.search("demo.s1")\n@@ -107,6 +105,7 @@\n assert re_suffix.search("demo.q1")\n assert re_suffix.search("demo.q1lk")\n \n+\n def clean_name(name):\n """Remove suffix."""\n match = re_suffix.search(name)\n@@ -117,6 +116,8 @@\n else:\n # Nothing to do\n return name\n+\n+\n assert clean_name("foo/1") == "foo"\n assert clean_name("foo/2") == "foo"\n assert clean_name("bar.f") == "bar"\n@@ -124,20 +125,22 @@\n assert clean_name("baz.p1") == "baz"\n assert clean_name("baz.q2") == "baz"\n \n-mapped_chars = { \'>\' :\'__gt__\',\n- \'<\' :\'__lt__\',\n- "\'" :\'__sq__\',\n- \'"\' :\'__dq__\',\n- \'[\' :\'__ob__\',\n- \']\' :\'__cb__\',\n- \'{\' :\'__oc__\',\n- \'}\' :\'__cc__\',\n- \'@\' : \'__at__\',\n- \'\\n\' : \'__cn__\',\n- \'\\r\' : \'__cr__\',\n- \'\\t\' : \'__tc__\',\n- \'#\' : \'__pd__\'\n- }\n+mapped_chars = {\n+ \'>\': \'__gt__\',\n+ \'<\': \'__lt__\',\n+ "\'": \'__sq__\',\n+ \'"\': \'__dq__\',\n+ \'[\': \'__ob__\',\n+ \']\': \'__cb__\',\n+ \'{\': \'__oc__\',\n+ \'}\': \'__cc__\',\n+ \'@\': \'__at__\',\n+ \'\\n\': \'__cn__\',\n+ \'\\r\': \'__cr__\',\n+ \'\\t\': \'__tc__\',\n+ \'#\': \'__pd__\',\n+}\n+\n \n def load_mapping_ids(filename, pair_mode, ids):\n """Parse SAM/BAM file, updating given set of ids.\n@@ -189,7 +192,7 @@\n if child.returncode:\n msg = "Error %i from \'samtools view %s\'\\n%s" % (child.returncode,\n filename, stderr)\n- sys_exit(msg.strip(), child.returncode)\n+ sys.exit(msg.strip(), child.returncode)\n else:\n handle.close()\n \n@@'..b'sitive_file is not None:\n+ elif pos_file is not None:\n print "Generating matching FASTQ file"\n- positive_handle = open(out_positive_file, "w")\n+ positive_handle = open(pos_file, "w")\n for title, seq, qual in FastqGeneralIterator(handle):\n- if clean_name(title.split(None, 1)[0]) in ids:\n+ if clean_name(title.split(None, 1)[0]) in wanted:\n positive_handle.write("@%s\\n%s\\n+\\n%s\\n" % (title, seq, qual))\n pos_count += 1\n else:\n neg_count += 1\n positive_handle.close()\n- elif out_negative_file is not None:\n+ elif neg_file is not None:\n print "Generating non-matching FASTQ file"\n- negative_handle = open(out_negative_file, "w")\n+ negative_handle = open(neg_file, "w")\n for title, seq, qual in FastqGeneralIterator(handle):\n- if clean_name(title.split(None, 1)[0]) in ids:\n+ if clean_name(title.split(None, 1)[0]) in wanted:\n pos_count += 1\n else:\n negative_handle.write("@%s\\n%s\\n+\\n%s\\n" % (title, seq, qual))\n@@ -328,48 +332,48 @@\n try:\n from Bio.SeqIO.SffIO import SffIterator, SffWriter\n except ImportError:\n- sys_exit("SFF filtering requires Biopython 1.54 or later")\n+ sys.exit("SFF filtering requires Biopython 1.54 or later")\n \n try:\n from Bio.SeqIO.SffIO import ReadRocheXmlManifest\n except ImportError:\n- #Prior to Biopython 1.56 this was a private function\n+ # Prior to Biopython 1.56 this was a private function\n from Bio.SeqIO.SffIO import _sff_read_roche_index_xml as ReadRocheXmlManifest\n \n- in_handle = open(in_file, "rb") #must be binary mode!\n+ in_handle = open(in_file, "rb") # must be binary mode!\n try:\n manifest = ReadRocheXmlManifest(in_handle)\n except ValueError:\n manifest = None\n \n- #This makes two passes though the SFF file with isn\'t so efficient,\n- #but this makes the code simple.\n+ # This makes two passes though the SFF file with isn\'t so efficient,\n+ # but this makes the code simple.\n pos_count = neg_count = 0\n- if out_positive_file is not None:\n- out_handle = open(out_positive_file, "wb")\n+ if pos_file is not None:\n+ out_handle = open(pos_file, "wb")\n writer = SffWriter(out_handle, xml=manifest)\n- in_handle.seek(0) #start again after getting manifest\n- pos_count = writer.write_file(rec for rec in SffIterator(in_handle) if clean_name(rec.id) in ids)\n+ in_handle.seek(0) # start again after getting manifest\n+ pos_count = writer.write_file(rec for rec in SffIterator(in_handle) if clean_name(rec.id) in wanted)\n out_handle.close()\n- if out_negative_file is not None:\n- out_handle = open(out_negative_file, "wb")\n+ if neg_file is not None:\n+ out_handle = open(neg_file, "wb")\n writer = SffWriter(out_handle, xml=manifest)\n- in_handle.seek(0) #start again\n- neg_count = writer.write_file(rec for rec in SffIterator(in_handle) if clean_name(rec.id) not in ids)\n+ in_handle.seek(0) # start again\n+ neg_count = writer.write_file(rec for rec in SffIterator(in_handle) if clean_name(rec.id) not in wanted)\n out_handle.close()\n- #And we\'re done\n+ # And we\'re done\n in_handle.close()\n return pos_count, neg_count\n \n \n-if seq_format.lower()=="sff":\n+if seq_format.lower() == "sff":\n sequence_filter = sff_filter\n-elif seq_format.lower()=="fasta":\n+elif seq_format.lower() == "fasta":\n sequence_filter = fasta_filter\n elif seq_format.lower().startswith("fastq"):\n sequence_filter = fastq_filter\n else:\n- sys_exit("Unsupported file type %r" % seq_format)\n+ sys.exit("Unsupported file type %r" % seq_format)\n \n pos_count, neg_count = sequence_filter(in_file, out_positive_file, out_negative_file, ids)\n print("%i mapped and %i unmapped reads." % (pos_count, neg_count))\n' |
| b |
| diff -r 8ff0ac66f1a3 -r 48e71dfd51b3 tools/seq_filter_by_mapping/seq_filter_by_mapping.xml --- a/tools/seq_filter_by_mapping/seq_filter_by_mapping.xml Wed May 13 11:08:58 2015 -0400 +++ b/tools/seq_filter_by_mapping/seq_filter_by_mapping.xml Wed May 10 13:16:44 2017 -0400 |
| b |
| @@ -1,28 +1,23 @@ -<tool id="seq_filter_by_mapping" name="Filter sequences by mapping" version="0.0.4"> +<tool id="seq_filter_by_mapping" name="Filter sequences by mapping" version="0.0.5"> <description>from SAM/BAM file</description> <requirements> - <requirement type="package" version="1.64">biopython</requirement> - <requirement type="python-module">Bio</requirement> - <requirement type="binary">samtools</requirement> + <requirement type="package" version="1.67">biopython</requirement> <requirement type="package" version="0.1.19">samtools</requirement> </requirements> - <stdio> - <!-- Anything other than zero is an error --> - <exit_code range="1:" /> - <exit_code range=":-1" /> - </stdio> - <version_command interpreter="python">seq_filter_by_mapping.py --version</version_command> - <command interpreter="python"> -seq_filter_by_mapping.py -i "$input_file" -f "$input_file.ext" -m $pair_mode + <version_command> +python $__tool_directory__/seq_filter_by_mapping.py --version + </version_command> + <command detect_errors="aggressive"> +python $__tool_directory__/seq_filter_by_mapping.py -i '$input_file' -f '$input_file.ext' -m $pair_mode #if $output_choice_cond.output_choice=="both" - -p $output_pos -n $output_neg + -p '$output_pos' -n '$output_neg' #elif $output_choice_cond.output_choice=="pos" - -p $output_pos + -p '$output_pos' #elif $output_choice_cond.output_choice=="neg" - -n $output_neg + -n '$output_neg' #end if ## Now loop over all the mapping files -#for i in $mapping_file#${i} #end for# +#for i in $mapping_file#'${i}' #end for# </command> <inputs> <param name="input_file" type="data" format="fasta,fastq,sff" label="Sequence file to be filtered" help="FASTA, FASTQ, or SFF format." /> |
| b |
| diff -r 8ff0ac66f1a3 -r 48e71dfd51b3 tools/seq_filter_by_mapping/tool_dependencies.xml --- a/tools/seq_filter_by_mapping/tool_dependencies.xml Wed May 13 11:08:58 2015 -0400 +++ b/tools/seq_filter_by_mapping/tool_dependencies.xml Wed May 10 13:16:44 2017 -0400 |
| b |
| @@ -1,9 +1,9 @@ <?xml version="1.0"?> <tool_dependency> - <package name="biopython" version="1.64"> - <repository changeset_revision="5477a05cc158" name="package_biopython_1_64" owner="biopython" toolshed="https://toolshed.g2.bx.psu.edu" /> + <package name="biopython" version="1.67"> + <repository changeset_revision="a42f244cce44" name="package_biopython_1_67" owner="biopython" toolshed="https://toolshed.g2.bx.psu.edu" /> </package> <package name="samtools" version="0.1.19"> - <repository changeset_revision="96aab723499f" name="package_samtools_0_1_19" owner="iuc" toolshed="https://toolshed.g2.bx.psu.edu" /> + <repository changeset_revision="c9bd782f5342" name="package_samtools_0_1_19" owner="iuc" toolshed="https://toolshed.g2.bx.psu.edu" /> </package> </tool_dependency> |