Previous changeset 3:5a3f5024ae07 (2017-10-12) Next changeset 5:f48f0dd2ed40 (2017-11-09) |
Commit message:
planemo upload |
added:
join_intervals_on_reference_alternative_splicing_events/UCSC_Main_on_Human_known_Alt_genome.bed join_intervals_on_reference_alternative_splicing_events/join.xml join_intervals_on_reference_alternative_splicing_events/macros.xml join_intervals_on_reference_alternative_splicing_events/operation_filter.py join_intervals_on_reference_alternative_splicing_events/test-data/1.bed join_intervals_on_reference_alternative_splicing_events/test-data/2.bed join_intervals_on_reference_transcipts/Reference_transcipts.interval join_intervals_on_reference_transcipts/join.xml join_intervals_on_reference_transcipts/macros.xml join_intervals_on_reference_transcipts/operation_filter.py join_intervals_on_reference_transcipts/test-data/1.bed join_intervals_on_reference_transcipts/test-data/2.bed |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_alternative_splicing_events/UCSC_Main_on_Human_known_Alt_genome.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_alternative_splicing_events/UCSC_Main_on_Human_known_Alt_genome.bed Sun Nov 05 20:41:54 2017 -0500 |
b |
b'@@ -0,0 +1,245965 @@\n+#chrom\tchromStart\tchromEnd\tname\n+chr1\t1048428\t1048578\taltPromoter\n+chr1\t1048528\t1048866\tbleedingExon\n+chr1\t44040071\t44040273\tbleedingExon\n+chr1\t155187538\t155191999\tstrangeSplice\n+chr1\t155189229\t155190079\tcassetteExon\n+chr1\t175097493\t175117015\tstrangeSplice\n+chr1\t197102647\t197143690\tstrangeSplice\n+chr1\t197132284\t197132352\tcassetteExon\n+chr1\t698959\t827670\tstrangeSplice\n+chr1\t711922\t827670\tstrangeSplice\n+chr1\t917486\t918021\tretainedIntron\n+chr1\t1179596\t1179746\taltPromoter\n+chr1\t1310688\t1310908\tretainedIntron\n+chr1\t13049949\t13149378\tstrangeSplice\n+chr1\t17694046\t17695157\tbleedingExon\n+chr1\t24379272\t24379652\tbleedingExon\n+chr1\t32374732\t32374882\taltPromoter\n+chr1\t39452532\t39452683\tbleedingExon\n+chr1\t40763362\t40763489\tcassetteExon\n+chr1\t42335595\t42337925\tstrangeSplice\n+chr1\t43383968\t43385045\tbleedingExon\n+chr1\t43384552\t43385045\tretainedIntron\n+chr1\t44412726\t44573001\tstrangeSplice\n+chr1\t45610553\t45613168\tbleedingExon\n+chr1\t45610995\t45613168\tretainedIntron\n+chr1\t50583664\t50705880\tstrangeSplice\n+chr1\t61901459\t62038473\tstrangeSplice\n+chr1\t62038530\t62162851\tstrangeSplice\n+chr1\t62128094\t62128457\tbleedingExon\n+chr1\t63568519\t63571020\tretainedIntron\n+chr1\t68024638\t68026573\tstrangeSplice\n+chr1\t70852868\t71012287\tstrangeSplice\n+chr1\t81306509\t81445035\tstrangeSplice\n+chr1\t83096977\t83160605\tstrangeSplice\n+chr1\t103546342\t103546976\tretainedIntron\n+chr1\t110880332\t110891391\tstrangeSplice\n+chr1\t117963938\t117966608\tatacIntron\n+chr1\t117964741\t117964810\tbleedingExon\n+chr1\t117964760\t117964910\taltPromoter\n+chr1\t150077190\t150077668\tbleedingExon\n+chr1\t150077231\t150077668\tretainedIntron\n+chr1\t150730195\t150732914\tbleedingExon\n+chr1\t150731120\t150732907\tretainedIntron\n+chr1\t150731134\t150732914\tretainedIntron\n+chr1\t152511578\t152600790\tstrangeSplice\n+chr1\t153616239\t153616389\taltPromoter\n+chr1\t153747363\t153748688\tretainedIntron\n+chr1\t155058138\t155058288\taltPromoter\n+chr1\t155320158\t155321624\tbleedingExon\n+chr1\t157993732\t158076112\tstrangeSplice\n+chr1\t160169672\t160171250\taltThreePrime\n+chr1\t164652139\t164831470\tstrangeSplice\n+chr1\t171700159\t171704386\tbleedingExon\n+chr1\t171702084\t171704386\tretainedIntron\n+chr1\t175528369\t175656604\tstrangeSplice\n+chr1\t183630637\t183633081\tretainedIntron\n+chr1\t193068915\t193069065\taltPromoter\n+chr1\t193069033\t193069634\tcassetteExon\n+chr1\t193069033\t193069181\tcassetteExon\n+chr1\t196342249\t196373139\tstrangeSplice\n+chr1\t196858408\t196912739\tstrangeSplice\n+chr1\t200406707\t200409560\tbleedingExon\n+chr1\t205913047\t205915360\tbleedingExon\n+chr1\t206037299\t206051852\tstrangeSplice\n+chr1\t206037299\t206051741\tstrangeSplice\n+chr1\t222943332\t222979622\tstrangeSplice\n+chr1\t226885537\t226885679\tcassetteExon\n+chr1\t236838518\t236850343\tstrangeSplice\n+chr1\t29453\t29603\taltPromoter\n+chr1\t30166\t30316\taltPromoter\n+chr1\t30266\t30563\tbleedingExon\n+chr1\t35244\t35276\tbleedingExon\n+chr1\t90050\t90286\tretainedIntron\n+chr1\t91055\t91205\taltPromoter\n+chr1\t110952\t110953\taltFinish\n+chr1\t120720\t120932\tcassetteExon\n+chr1\t120724\t120725\taltFinish\n+chr1\t120869\t120873\tstrangeSplice\n+chr1\t120869\t120873\tretainedIntron\n+chr1\t120873\t120932\tcassetteExon\n+chr1\t120882\t121032\taltPromoter\n+chr1\t133673\t133823\taltPromoter\n+chr1\t146509\t149707\tbleedingExon\n+chr1\t146509\t146641\tretainedIntron\n+chr1\t146781\t146931\taltPromoter\n+chr1\t149657\t149807\taltPromoter\n+chr1\t168717\t168867\taltPromoter\n+chr1\t173812\t173962\taltPromoter\n+chr1\t184971\t184976\tstrangeSplice\n+chr1\t184971\t184976\tretainedIntron\n+chr1\t185490\t185528\taltFivePrime\n+chr1\t186316\t186469\tcassetteExon\n+chr1\t187128\t187287\tcassetteExon\n+chr1\t187128\t187267\tcassetteExon\n+chr1\t187267\t187269\tstrangeSplice\n+chr1\t187267\t187269\tretainedIntron\n+chr1\t187269\t187287\tcassetteExon\n+chr1\t187375\t187379\taltFivePrime\n+chr1\t187886\t188125\tstrangeSplice\n+chr1\t188021\t188028\tcassetteExon\n+chr1\t188101\t188105\tcassetteExon\n+chr1\t188125\t188129\taltFivePrime\n+chr1\t188486\t188488\tstrangeSplice\n+chr1\t188889\t188902\taltThreePrime\n+chr1\t188889\t188892\taltThreePrime\n+chr1\t188892\t188902\taltThreePrime\n+chr1\t195361\t195511\taltPromoter\n+chr1\t200272\t200422\taltPromoter\n+chr1\t257863\t257864\t'..b'KI270928v1_alt\t46148\t46390\tretainedIntron\n+chr22_KI270928v1_alt\t46474\t46863\tretainedIntron\n+chr22_KI270928v1_alt\t46474\t46820\tretainedIntron\n+chr22_KI270928v1_alt\t46820\t46863\taltThreePrime\n+chr22_KI270928v1_alt\t50275\t50324\taltThreePrime\n+chr22_KI270928v1_alt\t51373\t51526\tcassetteExon\n+chr22_KI270928v1_alt\t52250\t53154\tbleedingExon\n+chr22_KI270928v1_alt\t52250\t52952\tretainedIntron\n+chr22_KI270928v1_alt\t53104\t53254\taltPromoter\n+chr22_KI270928v1_alt\t53197\t53347\taltPromoter\n+chr22_KI270928v1_alt\t54276\t54426\taltPromoter\n+chr22_KI270928v1_alt\t54715\t54948\tcassetteExon\n+chr22_KI270928v1_alt\t55385\t55436\tcassetteExon\n+chr22_KI270928v1_alt\t55675\t55756\tcassetteExon\n+chr22_KI270928v1_alt\t57482\t57587\tcassetteExon\n+chr22_KI270928v1_alt\t58025\t58091\tcassetteExon\n+chr22_KI270928v1_alt\t58630\t59041\tbleedingExon\n+chr22_KI270928v1_alt\t58630\t59019\tretainedIntron\n+chr22_KI270928v1_alt\t59397\t59398\taltFinish\n+chr22_KI270928v1_alt\t63590\t63882\tbleedingExon\n+chr22_KI270928v1_alt\t63688\t63973\tretainedIntron\n+chr22_KI270928v1_alt\t63688\t63882\tretainedIntron\n+chr22_KI270928v1_alt\t63688\t63716\taltThreePrime\n+chr22_KI270928v1_alt\t63716\t63719\tstrangeSplice\n+chr22_KI270928v1_alt\t63716\t63719\tretainedIntron\n+chr22_KI270928v1_alt\t63719\t63745\tcassetteExon\n+chr22_KI270928v1_alt\t63745\t63882\tretainedIntron\n+chr22_KI270928v1_alt\t63882\t63973\taltFivePrime\n+chr22_KI270928v1_alt\t65067\t65220\tcassetteExon\n+chr22_KI270928v1_alt\t65623\t65749\tbleedingExon\n+chr22_KI270928v1_alt\t66289\t66439\taltPromoter\n+chr22_KI270928v1_alt\t66668\t66889\tbleedingExon\n+chr22_KI270928v1_alt\t66668\t66669\tstrangeSplice\n+chr22_KI270928v1_alt\t66668\t66669\tretainedIntron\n+chr22_KI270928v1_alt\t66839\t66989\taltPromoter\n+chr22_KI270928v1_alt\t66867\t67017\taltPromoter\n+chr22_KI270928v1_alt\t90953\t91081\tcassetteExon\n+chr22_KI270928v1_alt\t90953\t91048\tcassetteExon\n+chr22_KI270928v1_alt\t91048\t91081\taltThreePrime\n+chr1_KI270711v1_random\t7930\t7934\tstrangeSplice\n+chr1_KI270711v1_random\t7930\t7934\tretainedIntron\n+chr1_KI270711v1_random\t10139\t10191\tcassetteExon\n+chr1_KI270711v1_random\t10910\t11083\tcassetteExon\n+chr1_KI270711v1_random\t12800\t12964\tcassetteExon\n+chr1_KI270711v1_random\t20006\t20221\tcassetteExon\n+chr1_KI270711v1_random\t21056\t21159\tcassetteExon\n+chr1_KI270711v1_random\t22932\t23142\tcassetteExon\n+chr1_KI270711v1_random\t22932\t23113\tcassetteExon\n+chr1_KI270711v1_random\t23113\t29459\tstrangeSplice\n+chr1_KI270711v1_random\t23113\t23142\taltThreePrime\n+chr1_KI270711v1_random\t24491\t24703\tcassetteExon\n+chr1_KI270711v1_random\t25166\t25239\tcassetteExon\n+chr1_KI270711v1_random\t26290\t26505\tcassetteExon\n+chr1_KI270711v1_random\t27336\t27439\tcassetteExon\n+chr1_KI270711v1_random\t29278\t29331\tcassetteExon\n+chr1_KI270711v1_random\t29331\t29333\tstrangeSplice\n+chr1_KI270711v1_random\t29331\t29333\tretainedIntron\n+chr1_KI270711v1_random\t29488\t29590\tretainedIntron\n+chr14_GL000194v1_random\t53589\t55445\tbleedingExon\n+chr14_GL000194v1_random\t53589\t53590\taltFinish\n+chr14_GL000194v1_random\t53593\t53594\taltFinish\n+chr14_GL000194v1_random\t54832\t55445\tretainedIntron\n+chr15_KI270727v1_random\t100257\t100260\tstrangeSplice\n+chr15_KI270727v1_random\t100310\t100726\tstrangeSplice\n+chr15_KI270727v1_random\t153237\t153458\tcassetteExon\n+chr16_KI270728v1_random\t1157757\t1240132\tstrangeSplice\n+chr16_KI270728v1_random\t1179378\t1257059\tstrangeSplice\n+chr16_KI270728v1_random\t1179522\t1209265\tstrangeSplice\n+chr16_KI270728v1_random\t1142043\t1142155\tstrangeSplice\n+chr16_KI270728v1_random\t1142711\t1142801\tstrangeSplice\n+chr16_KI270728v1_random\t1144087\t1144090\tstrangeSplice\n+chr16_KI270728v1_random\t1146604\t1146754\taltPromoter\n+chr16_KI270728v1_random\t1147427\t1147431\tstrangeSplice\n+chr16_KI270728v1_random\t1147818\t1147968\taltPromoter\n+chr16_KI270728v1_random\t1167356\t1167506\taltPromoter\n+chr16_KI270728v1_random\t1179190\t1179340\taltPromoter\n+chr16_KI270728v1_random\t1179250\t1179251\tstrangeSplice\n+chr16_KI270728v1_random\t1240187\t1240190\tstrangeSplice\n+chr17_GL000205v2_random\t100377\t100379\tstrangeSplice\n+chr17_GL000205v2_random\t100541\t100543\tstrangeSplice\n+chr22_KI270734v1_random\t161313\t161626\tcassetteExon\n' |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_alternative_splicing_events/join.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_alternative_splicing_events/join.xml Sun Nov 05 20:41:54 2017 -0500 |
[ |
@@ -0,0 +1,43 @@ +<tool id="join_ted_irase" name="Join_TED_IRASE" version="1.0.0"> + <description>Join Intervals on Reference Alternative Splicing Events</description> + <macros> + <import>macros.xml</import> + </macros> + <expand macro="requirements" /> + <code file="operation_filter.py"/> + <command><![CDATA[ +python '$__tool_directory__/join.py' +'$input1' +'$__tool_directory__/UCSC_Main_on_Human_known_Alt_genome.bed' +'$output' +-1 ${input1.metadata.chromCol},${input1.metadata.startCol},${input1.metadata.endCol},${input1.metadata.strandCol} +-2 ${input1.metadata.chromCol},${input1.metadata.startCol},${input1.metadata.endCol},${input1.metadata.strandCol} +-m $min +-f $fill + ]]></command> + <inputs> + <param name="input1" type="data" format="interval" label="Join" help="First dataset" /> + <param name="min" type="integer" value="1" label="With min overlap" help="(bp)" /> + <param name="fill" type="select" label="Return"> + <option value="none">Only records that are joined (INNER JOIN)</option> + <option value="right">All records of first dataset (fill null with ".")</option> + <option value="left">All records of second dataset (fill null with ".")</option> + <option value="both">All records of both datasets (fill nulls with ".")</option> + </param> + </inputs> + <outputs> + <data name="output" format_source="input1" metadata_source="input1" /> + </outputs> + <tests> + <test> + <param name="input1" value="1.bed" /> + <param name="input2" value="2.bed" /> + <param name="min" value="1" /> + <param name="fill" value="none" /> + <output name="output" file="output.dat" /> + </test> + </tests> + <help><![CDATA[ + Join Intervals on Reference Alternative Splicing Events + ]]></help> +</tool> |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_alternative_splicing_events/macros.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_alternative_splicing_events/macros.xml Sun Nov 05 20:41:54 2017 -0500 |
b |
@@ -0,0 +1,20 @@ +<?xml version="1.0"?> +<macros> + <xml name="requirements"> + <requirements> + <requirement type="package" version="0.7.1">bx-python</requirement> + <requirement type="package" version="1.0.0">galaxy-ops</requirement> + </requirements> + </xml> + <token name="@SCREENCASTS@"> +----- + +**Screencasts!** + +See Galaxy Interval Operation Screencasts_ (right click to open this link in another window). + +.. _Screencasts: https://galaxyproject.org/learn/interval-operations/ + +----- + </token> +</macros> |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_alternative_splicing_events/operation_filter.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_alternative_splicing_events/operation_filter.py Sun Nov 05 20:41:54 2017 -0500 |
[ |
@@ -0,0 +1,81 @@ +# runs after the job (and after the default post-filter) +from galaxy.jobs.handler import JOB_ERROR +from galaxy.tools.parameters import DataToolParameter + +# Older py compatibility +try: + set() +except: + from sets import Set as set + + +def validate_input( trans, error_map, param_values, page_param_map ): + dbkeys = set() + data_param_names = set() + data_params = 0 + for name, param in page_param_map.items(): + if isinstance( param, DataToolParameter ): + # for each dataset parameter + if param_values.get(name, None) is not None: + dbkeys.add( param_values[name].dbkey ) + data_params += 1 + # check meta data + try: + param = param_values[name] + if isinstance( param.datatype, trans.app.datatypes_registry.get_datatype_by_extension( 'gff' ).__class__ ): + # TODO: currently cannot validate GFF inputs b/c they are not derived from interval. + pass + else: # Validate interval datatype. + int( param.metadata.startCol ) + int( param.metadata.endCol ) + int( param.metadata.chromCol ) + if param.metadata.strandCol is not None: + int( param.metadata.strandCol ) + except: + error_msg = "The attributes of this dataset are not properly set. " + \ + "Click the pencil icon in the history item to set the chrom, start, end and strand columns." + error_map[name] = error_msg + data_param_names.add( name ) + if len( dbkeys ) > 1: + for name in data_param_names: + error_map[name] = "All datasets must belong to same genomic build, " \ + "this dataset is linked to build '%s'" % param_values[name].dbkey + if data_params != len(data_param_names): + for name in data_param_names: + error_map[name] = "A dataset of the appropriate type is required" + + +# Commented out by INS, 5/30/2007. What is the PURPOSE of this? +def exec_after_process(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + """Verify the output data after each run""" + for data in out_data.values(): + try: + if stderr and len( stderr ) > 0: + raise Exception( stderr ) + except Exception: + data.blurb = JOB_ERROR + data.state = JOB_ERROR + + +def exec_after_merge(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + exec_after_process( + app, inp_data, out_data, param_dict, tool=tool, stdout=stdout, stderr=stderr) + + # strip strand column if clusters were merged + for data in out_data.values(): + if param_dict['returntype'] is True: + data.metadata.chromCol = 1 + data.metadata.startCol = 2 + data.metadata.endCol = 3 + # merge always clobbers strand + data.metadata.strandCol = None + + +def exec_after_cluster(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + exec_after_process( + app, inp_data, out_data, param_dict, tool=tool, stdout=stdout, stderr=stderr) + + # strip strand column if clusters were merged + if param_dict["returntype"] == '1': + for data in out_data.values(): + data.metadata.strandCol = None |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_alternative_splicing_events/test-data/1.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_alternative_splicing_events/test-data/1.bed Sun Nov 05 20:41:54 2017 -0500 |
b |
@@ -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 - |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_alternative_splicing_events/test-data/2.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_alternative_splicing_events/test-data/2.bed Sun Nov 05 20:41:54 2017 -0500 |
b |
@@ -0,0 +1,68 @@ +chr1 147962192 147962580 NM_005997_cds_0_0_chr1_147962193_r 0 - +chr1 147984545 147984630 BC007833_cds_0_0_chr1_147984546_f 0 + +chr1 148078400 148078582 AJ011123_cds_0_0_chr1_148078401_r 0 - +chr1 148185136 148185276 NM_002796_cds_0_0_chr1_148185137_f 0 + +chr10 55251623 55253124 AY029205_cds_0_0_chr10_55251624_r 0 - +chr11 116124407 116124501 AK057832_cds_0_0_chr11_116124408_r 0 - +chr11 116206508 116206563 NM_000040_cds_1_0_chr11_116206509_f 0 + +chr11 116211733 116212337 BC005380_cds_0_0_chr11_116211734_r 0 - +chr11 130745911 130745993 AY358331_cds_0_0_chr11_130745912_f 0 + +chr12 38440094 38440321 NM_052885_cds_0_0_chr12_38440095_r 0 - +chr12 38905200 38905351 AY792511_cds_0_0_chr12_38905201_f 0 + +chr13 112381694 112381953 NM_207440_cds_1_0_chr13_112381695_f 0 + +chr13 29680676 29680875 NM_032116_cds_0_0_chr13_29680677_r 0 - +chr14 98521864 98521922 U88895_cds_0_0_chr14_98521865_f 0 + +chr14 98710240 98712285 NM_022898_cds_0_0_chr14_98710241_r 0 - +chr15 41486872 41487060 BX537418_cds_0_0_chr15_41486873_r 0 - +chr15 41673708 41673857 AK223365_cds_0_0_chr15_41673709_f 0 + +chr15 41679161 41679250 NM_153700_cds_0_0_chr15_41679162_r 0 - +chr15 41773540 41773689 AK223365_cds_0_0_chr15_41773541_f 0 + +chr16 142908 143003 NM_005332_cds_0_0_chr16_142909_f 0 + +chr16 179197 179339 BC065198_cds_0_0_chr16_179198_r 0 - +chr16 244413 244681 AK057165_cds_2_0_chr16_244414_f 0 + +chr16 259268 259383 AB016929_cds_0_0_chr16_259269_r 0 - +chr18 23786114 23786321 NM_001792_cds_0_0_chr18_23786115_r 0 - +chr18 59406881 59407046 NM_012397_cds_1_0_chr18_59406882_f 0 + +chr18 59455932 59456337 AB046400_cds_0_0_chr18_59455933_r 0 - +chr18 59528407 59528575 AY792326_cds_0_0_chr18_59528408_f 0 + +chr19 59068595 59069564 BC013995_cds_1_0_chr19_59068596_f 0 + +chr19 59236026 59236146 NM_198481_cds_0_0_chr19_59236027_r 0 - +chr19 59297998 59298008 NM_004542_cds_0_0_chr19_59297999_f 0 + +chr19 59318205 59318718 AK128544_cds_3_0_chr19_59318206_r 0 - +chr2 118288583 118288668 NM_006773_cds_0_0_chr2_118288584_f 0 + +chr2 118390395 118390500 BC005078_cds_0_0_chr2_118390396_r 0 - +chr2 220108689 220109267 AY125465_cds_0_0_chr2_220108690_f 0 + +chr2 220229609 220230869 NM_024536_cds_0_0_chr2_220229610_r 0 - +chr20 33330413 33330423 NM_181466_cds_0_0_chr20_33330414_r 0 - +chr20 33485370 33486123 BC085019_cds_1_0_chr20_33485371_f 0 + +chr20 33488491 33489122 NM_000557_cds_1_0_chr20_33488492_r 0 - +chr20 33513606 33513792 AF022655_cds_1_0_chr20_33513607_f 0 + +chr21 32687402 32687588 NM_032910_cds_0_0_chr21_32687403_f 0 + +chr21 32869641 32870022 NM_018277_cds_3_0_chr21_32869642_r 0 - +chr21 33321040 33322012 NM_005806_cds_1_0_chr21_33321041_f 0 + +chr21 33728358 33728724 AK129657_cds_0_0_chr21_33728359_r 0 - +chr22 30120223 30120265 NM_004147_cds_0_0_chr22_30120224_f 0 + +chr22 30160419 30160661 BC032941_cds_0_0_chr22_30160420_r 0 - +chr22 30228824 30228916 NM_001007467_cds_1_0_chr22_30228825_f 0 + +chr22 30340151 30340376 CR456540_cds_0_0_chr22_30340152_r 0 - +chr5 131311206 131311254 AF099740_cds_11_0_chr5_131311207_r 0 - +chr5 131424298 131424460 NM_000588_cds_0_0_chr5_131424299_f 0 + +chr5 131556601 131556672 BC035813_cds_0_0_chr5_131556602_r 0 - +chr5 131621326 131621419 BC003096_cds_0_0_chr5_131621327_f 0 + +chr6 108299600 108299744 NM_007214_cds_0_0_chr6_108299601_r 0 - +chr6 108594662 108594687 NM_003269_cds_0_0_chr6_108594663_f 0 + +chr6 108640045 108640151 NM_003795_cds_0_0_chr6_108640046_r 0 - +chr6 108722976 108723115 NM_145315_cds_0_0_chr6_108722977_f 0 + +chr7 113660517 113660685 AF467257_cds_1_0_chr7_113660518_f 0 + +chr7 116512159 116512389 NM_003391_cds_0_0_chr7_116512160_r 0 - +chr7 116714099 116714152 NM_000492_cds_0_0_chr7_116714100_f 0 + +chr7 116945541 116945787 AF377960_cds_0_0_chr7_116945542_r 0 - +chr8 118881131 118881317 NM_000127_cds_0_0_chr8_118881132_r 0 - +chr9 128764156 128764189 BC051300_cds_0_0_chr9_128764157_f 0 + +chr9 128787519 128789136 NM_014908_cds_0_0_chr9_128787520_r 0 - +chr9 128789552 128789584 NM_015354_cds_0_0_chr9_128789553_f 0 + +chr9 128850516 128850624 AB058751_cds_0_0_chr9_128850517_r 0 - +chrX 122745047 122745924 NM_001167_cds_1_0_chrX_122745048_f 0 + +chrX 152648964 152649196 NM_000425_cds_0_0_chrX_152648965_r 0 - +chrX 152691446 152691471 AF101728_cds_0_0_chrX_152691447_f 0 + +chrX 152694029 152694263 BC052303_cds_0_0_chrX_152694030_r 0 - |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_transcipts/Reference_transcipts.interval --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_transcipts/Reference_transcipts.interval Sun Nov 05 20:41:54 2017 -0500 |
b |
b'@@ -0,0 +1,74155 @@\n+chr1\t67092175\t67134971\t10_exon(s):C1orf141\n+chr1\t201283451\t201332993\t15_exon(s):PKP1\n+chr1\t67092175\t67134971\t9_exon(s):C1orf141\n+chr1\t67092175\t67134971\t8_exon(s):C1orf141\n+chr1\t201283451\t201332993\t14_exon(s):PKP1\n+chr1\t41847188\t42035925\t4_exon(s):HIVEP3\n+chr1\t33513998\t34165274\t71_exon(s):CSMD2\n+chr1\t8352403\t8817640\t24_exon(s):RERE\n+chr1\t33513998\t34165842\t70_exon(s):CSMD2\n+chr1\t8352403\t8423687\t13_exon(s):RERE\n+chr1\t8352403\t8817640\t23_exon(s):RERE\n+chr1\t83865031\t83999187\t22_exon(s):TTLL7\n+chr1\t83865031\t83999187\t20_exon(s):TTLL7\n+chr1\t83865031\t83999187\t21_exon(s):TTLL7\n+chr1\t75206131\t75616009\t22_exon(s):SLC44A5\n+chr1\t75202132\t75611116\t26_exon(s):SLC44A5\n+chr1\t92246397\t92299009\t19_exon(s):GLMN\n+chr1\t75202130\t75611114\t24_exon(s):SLC44A5\n+chr1\t92246397\t92299009\t18_exon(s):GLMN\n+chr1\t75202132\t75611116\t25_exon(s):SLC44A5\n+chr1\t75206389\t75611114\t24_exon(s):SLC44A5\n+chr1\t117367432\t117528870\t13_exon(s):MAN1A2\n+chr1\t92246397\t92299009\t18_exon(s):GLMN\n+chr1\t167721949\t167791919\t3_exon(s):MPZL1\n+chr1\t150982022\t150995638\t14_exon(s):ANXA9\n+chr1\t192935743\t192948257\t3_exon(s):LINC01032\n+chr1\t167721949\t167791919\t6_exon(s):MPZL1\n+chr1\t175944825\t176207244\t20_exon(s):RFWD2\n+chr1\t167721949\t167791919\t5_exon(s):MPZL1\n+chr1\t175944825\t176207244\t18_exon(s):RFWD2\n+chr1\t184387015\t184629021\t6_exon(s):C1orf21\n+chr1\t175944825\t176207244\t19_exon(s):RFWD2\n+chr1\t209661358\t209724125\t3_exon(s):LOC101930114\n+chr1\t209661358\t209742562\t3_exon(s):LOC101930114\n+chr1\t209686179\t209734950\t7_exon(s):HSD11B1\n+chr1\t243256004\t243500092\t20_exon(s):SDCCAG8\n+chr1\t243256004\t243500092\t19_exon(s):SDCCAG8\n+chr1\t243256004\t243500092\t19_exon(s):SDCCAG8\n+chr1\t243256004\t243500092\t18_exon(s):SDCCAG8\n+chr1\t243256004\t243500092\t21_exon(s):SDCCAG8\n+chr1\t243256004\t243500092\t18_exon(s):SDCCAG8\n+chr1\t209704790\t209734950\t6_exon(s):HSD11B1\n+chr1\t209686179\t209734950\t7_exon(s):HSD11B1\n+chr1\t2073461\t2185399\t16_exon(s):PRKCZ\n+chr1\t6785323\t7769706\t23_exon(s):CAMTA1\n+chr1\t2073461\t2185399\t16_exon(s):PRKCZ\n+chr1\t1020122\t1056119\t36_exon(s):AGRN\n+chr1\t3069177\t3438621\t17_exon(s):PRDM16\n+chr1\t3069177\t3438621\t17_exon(s):PRDM16\n+chr1\t6785323\t7769706\t23_exon(s):CAMTA1\n+chr1\t2073461\t2185399\t15_exon(s):PRKCZ\n+chr1\t2073461\t2185399\t15_exon(s):PRKCZ\n+chr1\t2073985\t2185399\t15_exon(s):PRKCZ\n+chr1\t1020122\t1056119\t39_exon(s):AGRN\n+chr1\t6264271\t6358944\t9_exon(s):ACOT7\n+chr1\t6264271\t6360704\t9_exon(s):ACOT7\n+chr1\t6264271\t6385823\t9_exon(s):ACOT7\n+chr1\t6785323\t7769706\t22_exon(s):CAMTA1\n+chr1\t6785323\t7769706\t23_exon(s):CAMTA1\n+chr1\t6785323\t7769706\t24_exon(s):CAMTA1\n+chr1\t6264271\t6393766\t9_exon(s):ACOT7\n+chr1\t2050469\t2185399\t18_exon(s):PRKCZ\n+chr1\t2073985\t2185399\t15_exon(s):PRKCZ\n+chr1\t2073461\t2185399\t15_exon(s):PRKCZ\n+chr1\t10474945\t10630761\t9_exon(s):PEX14\n+chr1\t14598716\t15118048\t15_exon(s):KAZN\n+chr1\t15684302\t15734769\t20_exon(s):PLEKHM2\n+chr1\t12567909\t12596396\t6_exon(s):DHRS3\n+chr1\t11479237\t11537583\t21_exon(s):DISP3\n+chr1\t12567909\t12616784\t6_exon(s):DHRS3\n+chr1\t12567909\t12617731\t6_exon(s):DHRS3\n+chr1\t14598716\t15068155\t8_exon(s):KAZN\n+chr1\t19920306\t19923617\t4_exon(s):PLA2G2E\n+chr1\t18871429\t18890891\t15_exon(s):ALDH4A1\n+chr1\t18871429\t18902799\t16_exon(s):ALDH4A1\n+chr1\t17755312\t17827063\t3_exon(s):ACTL8\n+chr1\t23019447\t23083691\t19_exon(s):KDM1A\n+chr1\t20806291\t21176888\t35_exon(s):EIF4G3\n+chr1\t20941244\t21168039\t13_exon(s):EIF4G3\n+chr1\t20806291\t21176888\t35_exon(s):EIF4G3\n+chr1\t18871429\t18902799\t15_exon(s):ALDH4A1\n+chr1\t18871429\t18902799\t14_exon(s):ALDH4A1\n+chr1\t20806291\t21111383\t32_exon(s):EIF4G3\n+chr1\t23019447\t23083691\t21_exon(s):KDM1A\n+chr1\t27234515\t27308633\t16_exon(s):WDTC1\n+chr1\t27234515\t27308633\t16_exon(s):WDTC1\n+chr1\t37715973\t37765152\t17_exon(s):EPHA10\n+chr1\t39841030\t39883511\t9_exon(s):TRIT1\n+chr1\t39841030\t39883511\t11_exon(s):TRIT1\n+chr1\t39841030\t39883511\t4_exon(s):TRIT1\n+chr1\t39841030\t39883511\t6_exon(s):TRIT1\n+chr1\t39841030\t39883511\t11_exon(s):TRIT1\n+chr1\t39841030\t39883511\t2_exon(s):TRIT1\n+chr1\t39841030\t39883511\t7_exon(s):TRIT1\n+chr1\t39841030\t39883511\t8_exon(s):TRIT1\n+chr1\t39841030\t39883511\t8_exon(s):TRIT1\n+chr1'..b'ndom\t50357\t53057\t5_exon(s):H19\n+chr11_KI270721v1_random\t51940\t52013\t1_exon(s):MIR675\n+chr11_KI270721v1_random\t50357\t53057\t5_exon(s):H19\n+chr14_GL000194v1_random\t53588\t115073\t5_exon(s):MAFIP\n+chr14_GL000194v1_random\t53588\t115073\t8_exon(s):MAFIP\n+chr14_GL000194v1_random\t53588\t115073\t3_exon(s):MAFIP\n+chr14_GL000194v1_random\t61659\t115073\t3_exon(s):MAFIP\n+chr15_KI270727v1_random\t241092\t241174\t1_exon(s):MIR5701-1\n+chr15_KI270727v1_random\t241092\t241174\t1_exon(s):MIR5701-2\n+chr15_KI270727v1_random\t241092\t241174\t1_exon(s):MIR5701-3\n+chr15_KI270727v1_random\t217529\t230123\t8_exon(s):NF1P2\n+chr15_KI270727v1_random\t136159\t167571\t11_exon(s):POTEB3\n+chr15_KI270727v1_random\t372321\t373405\t1_exon(s):OR4M2\n+chr16_KI270728v1_random\t17231\t19833\t2_exon(s):TP53TG3B\n+chr16_KI270728v1_random\t17231\t19833\t2_exon(s):TP53TG3\n+chr16_KI270728v1_random\t17231\t19833\t4_exon(s):TP53TG3\n+chr16_KI270728v1_random\t17231\t19833\t4_exon(s):TP53TG3B\n+chr16_KI270728v1_random\t17231\t19833\t3_exon(s):TP53TG3\n+chr16_KI270728v1_random\t17231\t19833\t3_exon(s):TP53TG3B\n+chr16_KI270728v1_random\t17233\t19273\t3_exon(s):TP53TG3C\n+chr16_KI270728v1_random\t17250\t19837\t2_exon(s):TP53TG3E\n+chr16_KI270728v1_random\t17251\t19833\t2_exon(s):TP53TG3F\n+chr16_KI270728v1_random\t17233\t19273\t2_exon(s):TP53TG3C\n+chr16_KI270728v1_random\t933855\t936466\t3_exon(s):TP53TG3D\n+chr16_KI270728v1_random\t1001629\t1037618\t25_exon(s):LOC102723753\n+chr16_KI270728v1_random\t1331814\t1346848\t3_exon(s):ENPP7P13\n+chr16_KI270728v1_random\t1331814\t1346848\t2_exon(s):ENPP7P13\n+chr17_GL000205v2_random\t54856\t57966\t1_exon(s):MGC70870\n+chr22_KI270731v1_random\t69141\t86923\t13_exon(s):GGT3P\n+chr22_KI270733v1_random\t122272\t135645\t1_exon(s):RNA45SN5\n+chr22_KI270733v1_random\t122272\t135645\t1_exon(s):RNA45SN1\n+chr22_KI270733v1_random\t122272\t135645\t1_exon(s):RNA45SN4\n+chr22_KI270733v1_random\t122272\t135645\t1_exon(s):RNA45SN2\n+chr22_KI270733v1_random\t130203\t135280\t1_exon(s):RNA28SN5\n+chr22_KI270733v1_random\t130203\t135280\t1_exon(s):RNA28SN1\n+chr22_KI270733v1_random\t130203\t135280\t1_exon(s):RNA28SN4\n+chr22_KI270733v1_random\t130203\t135280\t1_exon(s):RNA28SN2\n+chr22_KI270733v1_random\t121580\t121672\t1_exon(s):MIR6724-1\n+chr22_KI270733v1_random\t121580\t121672\t1_exon(s):MIR6724-2\n+chr22_KI270733v1_random\t121580\t121672\t1_exon(s):MIR6724-3\n+chr22_KI270733v1_random\t121580\t121672\t1_exon(s):MIR6724-4\n+chr22_KI270733v1_random\t125128\t125189\t1_exon(s):MIR3687-1\n+chr22_KI270733v1_random\t125128\t125189\t1_exon(s):MIR3687-2\n+chr22_KI270733v1_random\t125930\t127799\t1_exon(s):RNA18SN5\n+chr22_KI270733v1_random\t125930\t127799\t1_exon(s):RNA18SN1\n+chr22_KI270733v1_random\t125930\t127799\t1_exon(s):RNA18SN4\n+chr22_KI270733v1_random\t125930\t127799\t1_exon(s):RNA18SN2\n+chr22_KI270733v1_random\t125930\t127799\t1_exon(s):RNA18SN3\n+chr22_KI270733v1_random\t128875\t129032\t1_exon(s):RNA5-8SN5\n+chr22_KI270733v1_random\t128875\t129032\t1_exon(s):RNA5-8SN1\n+chr22_KI270733v1_random\t128875\t129032\t1_exon(s):RNA5-8SN2\n+chr22_KI270733v1_random\t128875\t129032\t1_exon(s):RNA5-8SN3\n+chr22_KI270733v1_random\t128876\t129032\t1_exon(s):RNA5-8SN4\n+chr22_KI270733v1_random\t170214\t170275\t1_exon(s):MIR3687-1\n+chr22_KI270733v1_random\t170214\t170275\t1_exon(s):MIR3687-2\n+chr22_KI270733v1_random\t171011\t172880\t1_exon(s):RNA18SN5\n+chr22_KI270733v1_random\t171011\t172880\t1_exon(s):RNA18SN1\n+chr22_KI270733v1_random\t171011\t172880\t1_exon(s):RNA18SN4\n+chr22_KI270733v1_random\t171011\t172880\t1_exon(s):RNA18SN2\n+chr22_KI270733v1_random\t171011\t172880\t1_exon(s):RNA18SN3\n+chr22_KI270733v1_random\t173954\t174111\t1_exon(s):RNA5-8SN5\n+chr22_KI270733v1_random\t173954\t174111\t1_exon(s):RNA5-8SN1\n+chr22_KI270733v1_random\t173954\t174111\t1_exon(s):RNA5-8SN2\n+chr22_KI270733v1_random\t173954\t174111\t1_exon(s):RNA5-8SN3\n+chr22_KI270733v1_random\t173955\t174111\t1_exon(s):RNA5-8SN4\n+chr22_KI270734v1_random\t72453\t74335\t1_exon(s):LOC102724728\n+chr22_KI270734v1_random\t90958\t98408\t4_exon(s):LOC105379550\n+chr22_KI270734v1_random\t138078\t161750\t14_exon(s):PRODH\n+chr22_KI270734v1_random\t138078\t161852\t15_exon(s):PRODH\n+chr22_KI270734v1_random\t131493\t137393\t5_exon(s):DGCR6\n' |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_transcipts/join.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_transcipts/join.xml Sun Nov 05 20:41:54 2017 -0500 |
[ |
@@ -0,0 +1,43 @@ +<tool id="join_ted_irt" name="Join_TED_IRT" version="1.0.0"> + <description>Join Intervals on Reference Transcipts</description> + <macros> + <import>macros.xml</import> + </macros> + <expand macro="requirements" /> + <code file="operation_filter.py"/> + <command><![CDATA[ +python '$__tool_directory__/join.py' +'$input1' +'$__tool_directory__/Reference_transcipts.interval' +'$output' +-1 ${input1.metadata.chromCol},${input1.metadata.startCol},${input1.metadata.endCol},${input1.metadata.strandCol} +-2 ${input1.metadata.chromCol},${input1.metadata.startCol},${input1.metadata.endCol},${input1.metadata.strandCol} +-m $min +-f $fill + ]]></command> + <inputs> + <param name="input1" type="data" format="interval" label="Join" help="First dataset" /> + <param name="min" type="integer" value="1" label="With min overlap" help="(bp)" /> + <param name="fill" type="select" label="Return"> + <option value="none">Only records that are joined (INNER JOIN)</option> + <option value="right">All records of first dataset (fill null with ".")</option> + <option value="left">All records of second dataset (fill null with ".")</option> + <option value="both">All records of both datasets (fill nulls with ".")</option> + </param> + </inputs> + <outputs> + <data name="output" format_source="input1" metadata_source="input1" /> + </outputs> + <tests> + <test> + <param name="input1" value="1.bed" /> + <param name="input2" value="2.bed" /> + <param name="min" value="1" /> + <param name="fill" value="none" /> + <output name="output" file="output.dat" /> + </test> + </tests> + <help><![CDATA[ + Join Intervals on Reference Transcipts + ]]></help> +</tool> |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_transcipts/macros.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_transcipts/macros.xml Sun Nov 05 20:41:54 2017 -0500 |
b |
@@ -0,0 +1,20 @@ +<?xml version="1.0"?> +<macros> + <xml name="requirements"> + <requirements> + <requirement type="package" version="0.7.1">bx-python</requirement> + <requirement type="package" version="1.0.0">galaxy-ops</requirement> + </requirements> + </xml> + <token name="@SCREENCASTS@"> +----- + +**Screencasts!** + +See Galaxy Interval Operation Screencasts_ (right click to open this link in another window). + +.. _Screencasts: https://galaxyproject.org/learn/interval-operations/ + +----- + </token> +</macros> |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_transcipts/operation_filter.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_transcipts/operation_filter.py Sun Nov 05 20:41:54 2017 -0500 |
[ |
@@ -0,0 +1,81 @@ +# runs after the job (and after the default post-filter) +from galaxy.jobs.handler import JOB_ERROR +from galaxy.tools.parameters import DataToolParameter + +# Older py compatibility +try: + set() +except: + from sets import Set as set + + +def validate_input( trans, error_map, param_values, page_param_map ): + dbkeys = set() + data_param_names = set() + data_params = 0 + for name, param in page_param_map.items(): + if isinstance( param, DataToolParameter ): + # for each dataset parameter + if param_values.get(name, None) is not None: + dbkeys.add( param_values[name].dbkey ) + data_params += 1 + # check meta data + try: + param = param_values[name] + if isinstance( param.datatype, trans.app.datatypes_registry.get_datatype_by_extension( 'gff' ).__class__ ): + # TODO: currently cannot validate GFF inputs b/c they are not derived from interval. + pass + else: # Validate interval datatype. + int( param.metadata.startCol ) + int( param.metadata.endCol ) + int( param.metadata.chromCol ) + if param.metadata.strandCol is not None: + int( param.metadata.strandCol ) + except: + error_msg = "The attributes of this dataset are not properly set. " + \ + "Click the pencil icon in the history item to set the chrom, start, end and strand columns." + error_map[name] = error_msg + data_param_names.add( name ) + if len( dbkeys ) > 1: + for name in data_param_names: + error_map[name] = "All datasets must belong to same genomic build, " \ + "this dataset is linked to build '%s'" % param_values[name].dbkey + if data_params != len(data_param_names): + for name in data_param_names: + error_map[name] = "A dataset of the appropriate type is required" + + +# Commented out by INS, 5/30/2007. What is the PURPOSE of this? +def exec_after_process(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + """Verify the output data after each run""" + for data in out_data.values(): + try: + if stderr and len( stderr ) > 0: + raise Exception( stderr ) + except Exception: + data.blurb = JOB_ERROR + data.state = JOB_ERROR + + +def exec_after_merge(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + exec_after_process( + app, inp_data, out_data, param_dict, tool=tool, stdout=stdout, stderr=stderr) + + # strip strand column if clusters were merged + for data in out_data.values(): + if param_dict['returntype'] is True: + data.metadata.chromCol = 1 + data.metadata.startCol = 2 + data.metadata.endCol = 3 + # merge always clobbers strand + data.metadata.strandCol = None + + +def exec_after_cluster(app, inp_data, out_data, param_dict, tool=None, stdout=None, stderr=None): + exec_after_process( + app, inp_data, out_data, param_dict, tool=tool, stdout=stdout, stderr=stderr) + + # strip strand column if clusters were merged + if param_dict["returntype"] == '1': + for data in out_data.values(): + data.metadata.strandCol = None |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_transcipts/test-data/1.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_transcipts/test-data/1.bed Sun Nov 05 20:41:54 2017 -0500 |
b |
@@ -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 - |
b |
diff -r 5a3f5024ae07 -r 4a501bd9d88c join_intervals_on_reference_transcipts/test-data/2.bed --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/join_intervals_on_reference_transcipts/test-data/2.bed Sun Nov 05 20:41:54 2017 -0500 |
b |
@@ -0,0 +1,68 @@ +chr1 147962192 147962580 NM_005997_cds_0_0_chr1_147962193_r 0 - +chr1 147984545 147984630 BC007833_cds_0_0_chr1_147984546_f 0 + +chr1 148078400 148078582 AJ011123_cds_0_0_chr1_148078401_r 0 - +chr1 148185136 148185276 NM_002796_cds_0_0_chr1_148185137_f 0 + +chr10 55251623 55253124 AY029205_cds_0_0_chr10_55251624_r 0 - +chr11 116124407 116124501 AK057832_cds_0_0_chr11_116124408_r 0 - +chr11 116206508 116206563 NM_000040_cds_1_0_chr11_116206509_f 0 + +chr11 116211733 116212337 BC005380_cds_0_0_chr11_116211734_r 0 - +chr11 130745911 130745993 AY358331_cds_0_0_chr11_130745912_f 0 + +chr12 38440094 38440321 NM_052885_cds_0_0_chr12_38440095_r 0 - +chr12 38905200 38905351 AY792511_cds_0_0_chr12_38905201_f 0 + +chr13 112381694 112381953 NM_207440_cds_1_0_chr13_112381695_f 0 + +chr13 29680676 29680875 NM_032116_cds_0_0_chr13_29680677_r 0 - +chr14 98521864 98521922 U88895_cds_0_0_chr14_98521865_f 0 + +chr14 98710240 98712285 NM_022898_cds_0_0_chr14_98710241_r 0 - +chr15 41486872 41487060 BX537418_cds_0_0_chr15_41486873_r 0 - +chr15 41673708 41673857 AK223365_cds_0_0_chr15_41673709_f 0 + +chr15 41679161 41679250 NM_153700_cds_0_0_chr15_41679162_r 0 - +chr15 41773540 41773689 AK223365_cds_0_0_chr15_41773541_f 0 + +chr16 142908 143003 NM_005332_cds_0_0_chr16_142909_f 0 + +chr16 179197 179339 BC065198_cds_0_0_chr16_179198_r 0 - +chr16 244413 244681 AK057165_cds_2_0_chr16_244414_f 0 + +chr16 259268 259383 AB016929_cds_0_0_chr16_259269_r 0 - +chr18 23786114 23786321 NM_001792_cds_0_0_chr18_23786115_r 0 - +chr18 59406881 59407046 NM_012397_cds_1_0_chr18_59406882_f 0 + +chr18 59455932 59456337 AB046400_cds_0_0_chr18_59455933_r 0 - +chr18 59528407 59528575 AY792326_cds_0_0_chr18_59528408_f 0 + +chr19 59068595 59069564 BC013995_cds_1_0_chr19_59068596_f 0 + +chr19 59236026 59236146 NM_198481_cds_0_0_chr19_59236027_r 0 - +chr19 59297998 59298008 NM_004542_cds_0_0_chr19_59297999_f 0 + +chr19 59318205 59318718 AK128544_cds_3_0_chr19_59318206_r 0 - +chr2 118288583 118288668 NM_006773_cds_0_0_chr2_118288584_f 0 + +chr2 118390395 118390500 BC005078_cds_0_0_chr2_118390396_r 0 - +chr2 220108689 220109267 AY125465_cds_0_0_chr2_220108690_f 0 + +chr2 220229609 220230869 NM_024536_cds_0_0_chr2_220229610_r 0 - +chr20 33330413 33330423 NM_181466_cds_0_0_chr20_33330414_r 0 - +chr20 33485370 33486123 BC085019_cds_1_0_chr20_33485371_f 0 + +chr20 33488491 33489122 NM_000557_cds_1_0_chr20_33488492_r 0 - +chr20 33513606 33513792 AF022655_cds_1_0_chr20_33513607_f 0 + +chr21 32687402 32687588 NM_032910_cds_0_0_chr21_32687403_f 0 + +chr21 32869641 32870022 NM_018277_cds_3_0_chr21_32869642_r 0 - +chr21 33321040 33322012 NM_005806_cds_1_0_chr21_33321041_f 0 + +chr21 33728358 33728724 AK129657_cds_0_0_chr21_33728359_r 0 - +chr22 30120223 30120265 NM_004147_cds_0_0_chr22_30120224_f 0 + +chr22 30160419 30160661 BC032941_cds_0_0_chr22_30160420_r 0 - +chr22 30228824 30228916 NM_001007467_cds_1_0_chr22_30228825_f 0 + +chr22 30340151 30340376 CR456540_cds_0_0_chr22_30340152_r 0 - +chr5 131311206 131311254 AF099740_cds_11_0_chr5_131311207_r 0 - +chr5 131424298 131424460 NM_000588_cds_0_0_chr5_131424299_f 0 + +chr5 131556601 131556672 BC035813_cds_0_0_chr5_131556602_r 0 - +chr5 131621326 131621419 BC003096_cds_0_0_chr5_131621327_f 0 + +chr6 108299600 108299744 NM_007214_cds_0_0_chr6_108299601_r 0 - +chr6 108594662 108594687 NM_003269_cds_0_0_chr6_108594663_f 0 + +chr6 108640045 108640151 NM_003795_cds_0_0_chr6_108640046_r 0 - +chr6 108722976 108723115 NM_145315_cds_0_0_chr6_108722977_f 0 + +chr7 113660517 113660685 AF467257_cds_1_0_chr7_113660518_f 0 + +chr7 116512159 116512389 NM_003391_cds_0_0_chr7_116512160_r 0 - +chr7 116714099 116714152 NM_000492_cds_0_0_chr7_116714100_f 0 + +chr7 116945541 116945787 AF377960_cds_0_0_chr7_116945542_r 0 - +chr8 118881131 118881317 NM_000127_cds_0_0_chr8_118881132_r 0 - +chr9 128764156 128764189 BC051300_cds_0_0_chr9_128764157_f 0 + +chr9 128787519 128789136 NM_014908_cds_0_0_chr9_128787520_r 0 - +chr9 128789552 128789584 NM_015354_cds_0_0_chr9_128789553_f 0 + +chr9 128850516 128850624 AB058751_cds_0_0_chr9_128850517_r 0 - +chrX 122745047 122745924 NM_001167_cds_1_0_chrX_122745048_f 0 + +chrX 152648964 152649196 NM_000425_cds_0_0_chrX_152648965_r 0 - +chrX 152691446 152691471 AF101728_cds_0_0_chrX_152691447_f 0 + +chrX 152694029 152694263 BC052303_cds_0_0_chrX_152694030_r 0 - |