Mercurial > repos > devteam > cut_columns
changeset 0:34c29e183ef7 draft
Uploaded cut_columns tarball.
| author | devteam |
|---|---|
| date | Tue, 04 Dec 2012 10:59:02 -0500 |
| parents | |
| children | 1378064e5d24 |
| files | cutWrapper.pl cutWrapper.xml test-data/1.bed test-data/eq-cut.dat |
| diffstat | 4 files changed, 409 insertions(+), 0 deletions(-) [+] |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/cutWrapper.pl Tue Dec 04 10:59:02 2012 -0500 @@ -0,0 +1,77 @@ +#!/usr/bin/perl -w + +use strict; +use warnings; + +my @columns = (); +my $del = ""; +my @in = (); +my @out = (); +my $command = ""; +my $field = 0; + +# a wrapper for cut for use in galaxy +# cutWrapper.pl [filename] [columns] [delim] [output] + +die "Check arguments\n" unless @ARGV == 4; + +$ARGV[1] =~ s/\s+//g; +foreach ( split /,/, $ARGV[1] ) { + if (m/^c\d{1,}$/i) { + push (@columns, $_); + $columns[@columns-1] =~s/c//ig; + } +} + +die "No columns specified, columns are not preceded with 'c', or commas are not used to separate column numbers: $ARGV[1]\n" if @columns == 0; + +my $column_delimiters_href = { + 'T' => q{\t}, + 'C' => ",", + 'D' => "-", + 'U' => "_", + 'P' => q{\|}, + 'Dt' => q{\.}, + 'Sp' => q{\s+} +}; + +$del = $column_delimiters_href->{$ARGV[2]}; + +open (OUT, ">$ARGV[3]") or die "Cannot create $ARGV[2]:$!\n"; +open (IN, "<$ARGV[0]") or die "Cannot open $ARGV[0]:$!\n"; + +while (my $line=<IN>) { + if ($line =~ /^#/) { + #Ignore comment lines + } else { + chop($line); + @in = split(/$del/, $line); + foreach $field (@columns) { + if (defined($in[$field-1])) { + push(@out, $in[$field-1]); + } else { + push(@out, "."); + } + } + print OUT join("\t",@out), "\n"; + @out = (); + } +} + +#while (<IN>) { +# chop; +# @in = split /$del/; +# foreach $field (@columns) { +# if (defined($in[$field-1])) { +# push(@out, $in[$field-1]); +# } else { +# push(@out, "."); +# } +# } +# print OUT join("\t",@out), "\n"; +# @out = (); +#} +close IN; + +close OUT; +
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/cutWrapper.xml Tue Dec 04 10:59:02 2012 -0500 @@ -0,0 +1,202 @@ +<tool id="Cut1" name="Cut" version="1.0.1"> + <description>columns from a table</description> + <command interpreter="perl">cutWrapper.pl $input "$columnList" $delimiter $out_file1</command> + <inputs> + <param name="columnList" size="10" type="text" value="c1,c2" label="Cut columns"/> + <param name="delimiter" type="select" label="Delimited by"> + <option value="T">Tab</option> + <option value="Sp">Whitespace</option> + <option value="Dt">Dot</option> + <option value="C">Comma</option> + <option value="D">Dash</option> + <option value="U">Underscore</option> + <option value="P">Pipe</option> + </param> + <param format="txt" name="input" type="data" label="From"/> + </inputs> + <outputs> + <data format="tabular" name="out_file1" > + <actions> + <conditional name="delimiter"> + <when value="T"> + <conditional name="input"> + <when datatype_isinstance="interval"> + <action type="format" default="tabular"> + <option type="from_param" name="columnList" column="0" offset="0"> <!-- chromCol is 1--> + + <filter type="insert_column" column="0" value="interval"/> + + <filter type="insert_column" ref="columnList" /> <!-- startCol --> + + <filter type="insert_column" ref="columnList" /> <!-- endCol --> + + <filter type="multiple_splitter" column="1" separator=","/> + <filter type="column_strip" column="1"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="1" name="lower" /> + <filter type="param_value" column="1" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="1" strip="c"/> <!-- get rid of c's --> + <filter type="boolean" column="1" cast="int" /> + + <filter type="multiple_splitter" column="2" separator=","/> + <filter type="column_strip" column="2"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="2" name="lower" /> + <filter type="param_value" column="2" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="2" strip="c"/> <!-- get rid of c's --> + <filter type="boolean" column="2" cast="int" /> + + <filter type="multiple_splitter" column="3" separator=","/> + <filter type="column_strip" column="3"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="3" name="lower" /> + <filter type="param_value" column="3" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="3" strip="c"/> <!-- get rid of c's --> + <filter type="boolean" column="3" cast="int" /> + + <filter type="metadata_value" ref="input" name="chromCol" column="1" /> + <filter type="metadata_value" ref="input" name="startCol" column="2" /> + <filter type="metadata_value" ref="input" name="endCol" column="3" /> + + </option> + </action> + + <conditional name="out_file1"> + <when datatype_isinstance="interval"> + <action type="metadata" name="chromCol"> + <option type="from_param" name="columnList" column="0" offset="0"> <!-- chromCol is 0--> + <filter type="multiple_splitter" column="0" separator=","/> + <filter type="column_strip" column="0"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="0" name="lower" /> + <filter type="param_value" column="0" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="0" strip="c"/> <!-- get rid of c's --> + <filter type="insert_column" value="1" iterate="True" column="0"/> + <filter type="boolean" column="1" cast="int" /> + <filter type="metadata_value" ref="input" name="chromCol" column="1" /> + </option> + </action> + + <action type="metadata" name="startCol"> + <option type="from_param" name="columnList" column="0" offset="0"> <!-- startCol is 0--> + <filter type="multiple_splitter" column="0" separator=","/> + <filter type="column_strip" column="0"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="0" name="lower" /> + <filter type="param_value" column="0" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="0" strip="c"/> <!-- get rid of c's --> + <filter type="insert_column" value="1" iterate="True" column="0"/> + <filter type="boolean" column="1" cast="int" /> + <filter type="metadata_value" ref="input" name="startCol" column="1" /> + </option> + </action> + + <action type="metadata" name="endCol"> + <option type="from_param" name="columnList" column="0" offset="0"> <!-- endCol is 0--> + <filter type="multiple_splitter" column="0" separator=","/> + <filter type="column_strip" column="0"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="0" name="lower" /> + <filter type="param_value" column="0" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="0" strip="c"/> <!-- get rid of c's --> + <filter type="insert_column" value="1" iterate="True" column="0"/> + <filter type="boolean" column="1" cast="int" /> + <filter type="metadata_value" ref="input" name="endCol" column="1" /> + </option> + </action> + + <action type="metadata" name="nameCol" default="0"> + <option type="from_param" name="columnList" column="0" offset="0"> <!-- nameCol is 0--> + <filter type="multiple_splitter" column="0" separator=","/> + <filter type="column_strip" column="0"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="0" name="lower" /> + <filter type="param_value" column="0" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="0" strip="c"/> <!-- get rid of c's --> + <filter type="insert_column" value="1" iterate="True" column="0"/> + <filter type="boolean" column="1" cast="int" /> + <filter type="metadata_value" ref="input" name="nameCol" column="1" /> + </option> + </action> + + <action type="metadata" name="strandCol" default="0"> + <option type="from_param" name="columnList" column="0" offset="0"> <!-- strandCol is 0--> + <filter type="multiple_splitter" column="0" separator=","/> + <filter type="column_strip" column="0"/> <!-- get rid of all external whitespace --> + <filter type="string_function" column="0" name="lower" /> + <filter type="param_value" column="0" value="^c\d{1,}$" compare="re_search" keep="True"/> + <filter type="column_strip" column="0" strip="c"/> <!-- get rid of c's --> + <filter type="insert_column" value="1" iterate="True" column="0"/> + <filter type="boolean" column="1" cast="int" /> + <filter type="metadata_value" ref="input" name="strandCol" column="1" /> + </option> + </action> + </when> + </conditional> + + </when> + </conditional> + </when> + </conditional> + </actions> + </data> + </outputs> + <tests> + <test> + <param name="columnList" value="c1,c4,c2,c3"/> + <param name="delimiter" value="T"/> + <param name="input" value="1.bed"/> + <output name="out_file1" file="eq-cut.dat"/> + </test> + </tests> + <help> + +.. class:: warningmark + +**WARNING: This tool breaks column assignments.** To re-establish column assignments run the tools and click on the pencil icon in the latest history item. + +.. class:: infomark + +The output of this tool is always in tabular format (e.g., if your original delimiters are commas, they will be replaced with tabs). For example: + + Cutting columns 1 and 3 from:: + + apple,is,good + windows,is,bad + + will give:: + + apple good + windows bad + +----- + +**What it does** + +This tool selects (cuts out) specified columns from the dataset. + +- Columns are specified as **c1**, **c2**, and so on. Column count begins with **1** +- Columns can be specified in any order (e.g., **c2,c1,c6**) +- If you specify more columns than actually present - empty spaces will be filled with dots + +----- + +**Example** + +Input dataset (six columns: c1, c2, c3, c4, c5, and c6):: + + chr1 10 1000 gene1 0 + + chr2 100 1500 gene2 0 + + +**cut** on columns "**c1,c4,c6**" will return:: + + chr1 gene1 + + chr2 gene2 + + +**cut** on columns "**c6,c5,c4,c1**" will return:: + + + 0 gene1 chr1 + + 0 gene2 chr2 + + +**cut** on columns "**c8,c7,c4**" will return:: + + . . gene1 + . . gene2 + + +</help> +</tool>
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/1.bed Tue Dec 04 10:59:02 2012 -0500 @@ -0,0 +1,65 @@ +chr1 147962192 147962580 CCDS989.1_cds_0_0_chr1_147962193_r 0 - +chr1 147984545 147984630 CCDS990.1_cds_0_0_chr1_147984546_f 0 + +chr1 148078400 148078582 CCDS993.1_cds_0_0_chr1_148078401_r 0 - +chr1 148185136 148185276 CCDS996.1_cds_0_0_chr1_148185137_f 0 + +chr10 55251623 55253124 CCDS7248.1_cds_0_0_chr10_55251624_r 0 - +chr11 116124407 116124501 CCDS8374.1_cds_0_0_chr11_116124408_r 0 - +chr11 116206508 116206563 CCDS8377.1_cds_0_0_chr11_116206509_f 0 + +chr11 116211733 116212337 CCDS8378.1_cds_0_0_chr11_116211734_r 0 - +chr11 1812377 1812407 CCDS7726.1_cds_0_0_chr11_1812378_f 0 + +chr12 38440094 38440321 CCDS8736.1_cds_0_0_chr12_38440095_r 0 - +chr13 112381694 112381953 CCDS9526.1_cds_0_0_chr13_112381695_f 0 + +chr14 98710240 98712285 CCDS9949.1_cds_0_0_chr14_98710241_r 0 - +chr15 41486872 41487060 CCDS10096.1_cds_0_0_chr15_41486873_r 0 - +chr15 41673708 41673857 CCDS10097.1_cds_0_0_chr15_41673709_f 0 + +chr15 41679161 41679250 CCDS10098.1_cds_0_0_chr15_41679162_r 0 - +chr15 41826029 41826196 CCDS10101.1_cds_0_0_chr15_41826030_f 0 + +chr16 142908 143003 CCDS10397.1_cds_0_0_chr16_142909_f 0 + +chr16 179963 180135 CCDS10401.1_cds_0_0_chr16_179964_r 0 - +chr16 244413 244681 CCDS10402.1_cds_0_0_chr16_244414_f 0 + +chr16 259268 259383 CCDS10403.1_cds_0_0_chr16_259269_r 0 - +chr18 23786114 23786321 CCDS11891.1_cds_0_0_chr18_23786115_r 0 - +chr18 59406881 59407046 CCDS11985.1_cds_0_0_chr18_59406882_f 0 + +chr18 59455932 59456337 CCDS11986.1_cds_0_0_chr18_59455933_r 0 - +chr18 59600586 59600754 CCDS11988.1_cds_0_0_chr18_59600587_f 0 + +chr19 59068595 59069564 CCDS12866.1_cds_0_0_chr19_59068596_f 0 + +chr19 59236026 59236146 CCDS12872.1_cds_0_0_chr19_59236027_r 0 - +chr19 59297998 59298008 CCDS12877.1_cds_0_0_chr19_59297999_f 0 + +chr19 59302168 59302288 CCDS12878.1_cds_0_0_chr19_59302169_r 0 - +chr2 118288583 118288668 CCDS2120.1_cds_0_0_chr2_118288584_f 0 + +chr2 118394148 118394202 CCDS2121.1_cds_0_0_chr2_118394149_r 0 - +chr2 220190202 220190242 CCDS2441.1_cds_0_0_chr2_220190203_f 0 + +chr2 220229609 220230869 CCDS2443.1_cds_0_0_chr2_220229610_r 0 - +chr20 33330413 33330423 CCDS13249.1_cds_0_0_chr20_33330414_r 0 - +chr20 33513606 33513792 CCDS13255.1_cds_0_0_chr20_33513607_f 0 + +chr20 33579500 33579527 CCDS13256.1_cds_0_0_chr20_33579501_r 0 - +chr20 33593260 33593348 CCDS13257.1_cds_0_0_chr20_33593261_f 0 + +chr21 32707032 32707192 CCDS13614.1_cds_0_0_chr21_32707033_f 0 + +chr21 32869641 32870022 CCDS13615.1_cds_0_0_chr21_32869642_r 0 - +chr21 33321040 33322012 CCDS13620.1_cds_0_0_chr21_33321041_f 0 + +chr21 33744994 33745040 CCDS13625.1_cds_0_0_chr21_33744995_r 0 - +chr22 30120223 30120265 CCDS13897.1_cds_0_0_chr22_30120224_f 0 + +chr22 30160419 30160661 CCDS13898.1_cds_0_0_chr22_30160420_r 0 - +chr22 30665273 30665360 CCDS13901.1_cds_0_0_chr22_30665274_f 0 + +chr22 30939054 30939266 CCDS13903.1_cds_0_0_chr22_30939055_r 0 - +chr5 131424298 131424460 CCDS4149.1_cds_0_0_chr5_131424299_f 0 + +chr5 131556601 131556672 CCDS4151.1_cds_0_0_chr5_131556602_r 0 - +chr5 131621326 131621419 CCDS4152.1_cds_0_0_chr5_131621327_f 0 + +chr5 131847541 131847666 CCDS4155.1_cds_0_0_chr5_131847542_r 0 - +chr6 108299600 108299744 CCDS5061.1_cds_0_0_chr6_108299601_r 0 - +chr6 108594662 108594687 CCDS5063.1_cds_0_0_chr6_108594663_f 0 + +chr6 108640045 108640151 CCDS5064.1_cds_0_0_chr6_108640046_r 0 - +chr6 108722976 108723115 CCDS5067.1_cds_0_0_chr6_108722977_f 0 + +chr7 113660517 113660685 CCDS5760.1_cds_0_0_chr7_113660518_f 0 + +chr7 116512159 116512389 CCDS5771.1_cds_0_0_chr7_116512160_r 0 - +chr7 116714099 116714152 CCDS5773.1_cds_0_0_chr7_116714100_f 0 + +chr7 116945541 116945787 CCDS5774.1_cds_0_0_chr7_116945542_r 0 - +chr8 118881131 118881317 CCDS6324.1_cds_0_0_chr8_118881132_r 0 - +chr9 128764156 128764189 CCDS6914.1_cds_0_0_chr9_128764157_f 0 + +chr9 128787519 128789136 CCDS6915.1_cds_0_0_chr9_128787520_r 0 - +chr9 128882427 128882523 CCDS6917.1_cds_0_0_chr9_128882428_f 0 + +chr9 128937229 128937445 CCDS6919.1_cds_0_0_chr9_128937230_r 0 - +chrX 122745047 122745924 CCDS14606.1_cds_0_0_chrX_122745048_f 0 + +chrX 152648964 152649196 CCDS14733.1_cds_0_0_chrX_152648965_r 0 - +chrX 152691446 152691471 CCDS14735.1_cds_0_0_chrX_152691447_f 0 + +chrX 152694029 152694263 CCDS14736.1_cds_0_0_chrX_152694030_r 0 -
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/eq-cut.dat Tue Dec 04 10:59:02 2012 -0500 @@ -0,0 +1,65 @@ +chr1 CCDS989.1_cds_0_0_chr1_147962193_r 147962192 147962580 +chr1 CCDS990.1_cds_0_0_chr1_147984546_f 147984545 147984630 +chr1 CCDS993.1_cds_0_0_chr1_148078401_r 148078400 148078582 +chr1 CCDS996.1_cds_0_0_chr1_148185137_f 148185136 148185276 +chr10 CCDS7248.1_cds_0_0_chr10_55251624_r 55251623 55253124 +chr11 CCDS8374.1_cds_0_0_chr11_116124408_r 116124407 116124501 +chr11 CCDS8377.1_cds_0_0_chr11_116206509_f 116206508 116206563 +chr11 CCDS8378.1_cds_0_0_chr11_116211734_r 116211733 116212337 +chr11 CCDS7726.1_cds_0_0_chr11_1812378_f 1812377 1812407 +chr12 CCDS8736.1_cds_0_0_chr12_38440095_r 38440094 38440321 +chr13 CCDS9526.1_cds_0_0_chr13_112381695_f 112381694 112381953 +chr14 CCDS9949.1_cds_0_0_chr14_98710241_r 98710240 98712285 +chr15 CCDS10096.1_cds_0_0_chr15_41486873_r 41486872 41487060 +chr15 CCDS10097.1_cds_0_0_chr15_41673709_f 41673708 41673857 +chr15 CCDS10098.1_cds_0_0_chr15_41679162_r 41679161 41679250 +chr15 CCDS10101.1_cds_0_0_chr15_41826030_f 41826029 41826196 +chr16 CCDS10397.1_cds_0_0_chr16_142909_f 142908 143003 +chr16 CCDS10401.1_cds_0_0_chr16_179964_r 179963 180135 +chr16 CCDS10402.1_cds_0_0_chr16_244414_f 244413 244681 +chr16 CCDS10403.1_cds_0_0_chr16_259269_r 259268 259383 +chr18 CCDS11891.1_cds_0_0_chr18_23786115_r 23786114 23786321 +chr18 CCDS11985.1_cds_0_0_chr18_59406882_f 59406881 59407046 +chr18 CCDS11986.1_cds_0_0_chr18_59455933_r 59455932 59456337 +chr18 CCDS11988.1_cds_0_0_chr18_59600587_f 59600586 59600754 +chr19 CCDS12866.1_cds_0_0_chr19_59068596_f 59068595 59069564 +chr19 CCDS12872.1_cds_0_0_chr19_59236027_r 59236026 59236146 +chr19 CCDS12877.1_cds_0_0_chr19_59297999_f 59297998 59298008 +chr19 CCDS12878.1_cds_0_0_chr19_59302169_r 59302168 59302288 +chr2 CCDS2120.1_cds_0_0_chr2_118288584_f 118288583 118288668 +chr2 CCDS2121.1_cds_0_0_chr2_118394149_r 118394148 118394202 +chr2 CCDS2441.1_cds_0_0_chr2_220190203_f 220190202 220190242 +chr2 CCDS2443.1_cds_0_0_chr2_220229610_r 220229609 220230869 +chr20 CCDS13249.1_cds_0_0_chr20_33330414_r 33330413 33330423 +chr20 CCDS13255.1_cds_0_0_chr20_33513607_f 33513606 33513792 +chr20 CCDS13256.1_cds_0_0_chr20_33579501_r 33579500 33579527 +chr20 CCDS13257.1_cds_0_0_chr20_33593261_f 33593260 33593348 +chr21 CCDS13614.1_cds_0_0_chr21_32707033_f 32707032 32707192 +chr21 CCDS13615.1_cds_0_0_chr21_32869642_r 32869641 32870022 +chr21 CCDS13620.1_cds_0_0_chr21_33321041_f 33321040 33322012 +chr21 CCDS13625.1_cds_0_0_chr21_33744995_r 33744994 33745040 +chr22 CCDS13897.1_cds_0_0_chr22_30120224_f 30120223 30120265 +chr22 CCDS13898.1_cds_0_0_chr22_30160420_r 30160419 30160661 +chr22 CCDS13901.1_cds_0_0_chr22_30665274_f 30665273 30665360 +chr22 CCDS13903.1_cds_0_0_chr22_30939055_r 30939054 30939266 +chr5 CCDS4149.1_cds_0_0_chr5_131424299_f 131424298 131424460 +chr5 CCDS4151.1_cds_0_0_chr5_131556602_r 131556601 131556672 +chr5 CCDS4152.1_cds_0_0_chr5_131621327_f 131621326 131621419 +chr5 CCDS4155.1_cds_0_0_chr5_131847542_r 131847541 131847666 +chr6 CCDS5061.1_cds_0_0_chr6_108299601_r 108299600 108299744 +chr6 CCDS5063.1_cds_0_0_chr6_108594663_f 108594662 108594687 +chr6 CCDS5064.1_cds_0_0_chr6_108640046_r 108640045 108640151 +chr6 CCDS5067.1_cds_0_0_chr6_108722977_f 108722976 108723115 +chr7 CCDS5760.1_cds_0_0_chr7_113660518_f 113660517 113660685 +chr7 CCDS5771.1_cds_0_0_chr7_116512160_r 116512159 116512389 +chr7 CCDS5773.1_cds_0_0_chr7_116714100_f 116714099 116714152 +chr7 CCDS5774.1_cds_0_0_chr7_116945542_r 116945541 116945787 +chr8 CCDS6324.1_cds_0_0_chr8_118881132_r 118881131 118881317 +chr9 CCDS6914.1_cds_0_0_chr9_128764157_f 128764156 128764189 +chr9 CCDS6915.1_cds_0_0_chr9_128787520_r 128787519 128789136 +chr9 CCDS6917.1_cds_0_0_chr9_128882428_f 128882427 128882523 +chr9 CCDS6919.1_cds_0_0_chr9_128937230_r 128937229 128937445 +chrX CCDS14606.1_cds_0_0_chrX_122745048_f 122745047 122745924 +chrX CCDS14733.1_cds_0_0_chrX_152648965_r 152648964 152649196 +chrX CCDS14735.1_cds_0_0_chrX_152691447_f 152691446 152691471 +chrX CCDS14736.1_cds_0_0_chrX_152694030_r 152694029 152694263