| Next changeset 1:ca3fcd18dc3e (2015-11-11) |
|
Commit message:
Uploaded tool tarball. |
|
added:
disease_ontology_gene_fuzzy_selector.pl funDo.xml test-data/funDo_output1.interval tool-data/funDo.loc.sample |
| b |
| diff -r 000000000000 -r 64133669255f disease_ontology_gene_fuzzy_selector.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/disease_ontology_gene_fuzzy_selector.pl Wed Sep 25 11:22:47 2013 -0400 |
| [ |
| @@ -0,0 +1,64 @@ +#!/usr/bin/env perl + +use strict; +use warnings; + +################################################################## +# Select genes that are associated with the diseases listed in the +# disease ontology. +# ontology: http://do-wiki.nubic.northwestern.edu/index.php/Main_Page +# gene associations by FunDO: http://projects.bioinformatics.northwestern.edu/do_rif/ +# Sept 2010, switch to doLite +# input: build outfile sourceFileLoc.loc term or partial term +################################################################## + +if (!@ARGV or @ARGV < 3) { + print "usage: disease_ontology_gene_selector.pl build outfile.txt sourceFile.loc [list of terms]\n"; + exit; +} + +my $build = shift @ARGV; +my $out = shift @ARGV; +my $in = shift @ARGV; +my $term = shift @ARGV; +$term =~ s/^'//; #remove quotes protecting from shell +$term =~ s/'$//; +my $data; +open(LOC, $in) or die "Couldn't open $in, $!\n"; +while (<LOC>) { + chomp; + if (/^\s*#/) { next; } + my @f = split(/\t/); + if ($f[0] eq $build) { + if ($f[1] eq 'disease associated genes') { + $data = $f[2]; + } + } +} +close LOC or die "Couldn't close $in, $!\n"; +if (!$data) { + print "Error $build not found in $in\n"; + exit; +} +if (!defined $term) { + print "No disease term entered\n"; + exit; +} + +#start with just fuzzy term matches +open(OUT, ">", $out) or die "Couldn't open $out, $!\n"; +open(FH, $data) or die "Couldn't open data file $data, $!\n"; +$term =~ s/\s+/|/g; #use OR between words +while (<FH>) { + chomp; + my @f = split(/\t/); #chrom start end strand geneName geneID disease + if ($f[6] =~ /($term)/i) { + print OUT join("\t", @f), "\n"; + }elsif ($term eq 'disease') { #print all with disease + print OUT join("\t", @f), "\n"; + } +} +close FH or die "Couldn't close data file $data, $!\n"; +close OUT or die "Couldn't close $out, $!\n"; + +exit; |
| b |
| diff -r 000000000000 -r 64133669255f funDo.xml --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/funDo.xml Wed Sep 25 11:22:47 2013 -0400 |
| b |
| @@ -0,0 +1,101 @@ +<tool id="hgv_funDo" name="FunDO" version="1.0.0"> + <description>human genes associated with disease terms</description> + + <command interpreter="perl"> + disease_ontology_gene_fuzzy_selector.pl $build $out_file1 ${GALAXY_DATA_INDEX_DIR}/funDo.loc '$term' + </command> + + <inputs> + <param name="build" type="select" label="Database build"> + <options from_file="funDo.loc"> + <column name="name" index="0"/> + <column name="value" index="0"/> + <filter type="unique_value" column="0"/> + </options> + </param> + <param name="term" size="40" type="text" label="Disease term(s)" /> + </inputs> + + <outputs> + <data format="interval" name="out_file1"> + </data> + </outputs> + + <tests> + <test> + <param name="term" value="lung"/> + <param name="build" value="hg18"/> + <output name="out_file1" file="funDo_output1.interval" /> + </test> + </tests> + + <help> +**Dataset formats** + +There is no input dataset. The output is in interval_ format. + +.. _interval: ${static_path}/formatHelp.html#interval + +----- + +**What it does** + +This tool searches the disease-term field of the DOLite mappings +used by the FunDO project and returns a set of genes that +are associated with terms matching the specified pattern. (This is the +reverse of what FunDO's own server does.) + +The search is case insensitive, and selects terms that contain any of +the given words, either exactly or within a longer word (e.g. "nemia" +selects not only "anemia", but also "hyperglycinemia", "tyrosinemias", +and many other things). Multiple words should be separated by spaces, +not commas. As a special case, entering the word "disease" returns all +genes associated with any disease, even if that word does not actually +appear in the term field. + +Website: http://django.nubic.northwestern.edu/fundo/ + +----- + +**Example** + +Typing:: + + carcinoma + +results in:: + + 1. 2. 3. 4. 5. 6. 7. + chr11 89507465 89565427 + NAALAD2 10003 Adenocarcinoma + chr15 50189113 50192264 - BCL2L10 10017 Carcinoma + chr7 150535855 150555250 - ABCF2 10061 Clear cell carcinoma + chr7 150540508 150555250 - ABCF2 10061 Clear cell carcinoma + chr10 134925911 134940397 - ADAM8 101 Adenocarcinoma + chr10 134925911 134940397 - ADAM8 101 Adenocarcinoma + etc. + +where the column contents are as follows:: + + 1. chromosome name + 2. start position of the gene + 3. end position of the gene + 4. strand + 4. gene name + 6. Entrez Gene ID + 7. disease term + +----- + +**References** + +Du P, Feng G, Flatow J, Song J, Holko M, Kibbe WA, Lin SM. (2009) +From disease ontology to disease-ontology lite: statistical methods to adapt a general-purpose +ontology for the test of gene-ontology associations. +Bioinformatics. 25(12):i63-8. + +Osborne JD, Flatow J, Holko M, Lin SM, Kibbe WA, Zhu LJ, Danila MI, Feng G, Chisholm RL. (2009) +Annotating the human genome with Disease Ontology. +BMC Genomics. 10 Suppl 1:S6. + + </help> +</tool> |
| b |
| diff -r 000000000000 -r 64133669255f test-data/funDo_output1.interval --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/funDo_output1.interval Wed Sep 25 11:22:47 2013 -0400 |
| b |
| b'@@ -0,0 +1,483 @@\n+chr5\t324735\t368089\t+\tPDCD6\t10016\tLung cancer\n+chr10\t134925911\t134940397\t-\tADAM8\t101\tLung cancer\n+chr10\t134925911\t134940397\t-\tADAM8\t101\tLung cancer\n+chr10\t134925911\t134940397\t-\tADAM8\t101\tLung cancer\n+chr17\t46294585\t46296412\t-\tTOB1\t10140\tLung cancer\n+chr3\t50101371\t50131396\t+\tRBM5\t10181\tLung cancer\n+chr13\t94546025\t94751688\t-\tABCC4\t10257\tLung cancer\n+chr13\t94470083\t94751688\t-\tABCC4\t10257\tLung cancer\n+chr17\t38166737\t38168585\t+\tRAMP2\t10266\tLung cancer\n+chr16\t22264757\t22293439\t-\tCDR2\t1039\tLung disease\n+chr1\t144287344\t144297903\t+\tPIAS3\t10401\tLung cancer\n+chr19\t38556448\t38565432\t+\tCEBPG\t1054\tLung cancer\n+chr16\t66153964\t66230589\t+\tCTCF\t10664\tLung cancer\n+chr1\t11009166\t11029872\t-\tMASP2\t10747\tInfectious lung disease\n+chr1\t11027441\t11029872\t-\tMASP2\t10747\tInfectious lung disease\n+chrX\t133527538\t133620179\t-\tPLAC1\t10761\tLung cancer\n+chr8\t24207524\t24268671\t+\tADAM28\t10863\tLung cancer\n+chr8\t24207524\t24249557\t+\tADAM28\t10863\tLung cancer\n+chr18\t9465529\t9528106\t+\tRALBP1\t10928\tLung cancer\n+chr2\t9641556\t9688557\t-\tYWHAQ\t10971\tLung cancer\n+chr14\t23875066\t23879082\t-\tRIPK3\t11035\tLung cancer\n+chr5\t54309452\t54317171\t-\tESM1\t11082\tLung cancer\n+chr5\t54309452\t54317171\t-\tESM1\t11082\tLung cancer\n+chr6\t167332805\t167374056\t+\tFGFR1OP\t11116\tLung cancer\n+chr6\t167332805\t167374056\t+\tFGFR1OP\t11116\tLung cancer\n+chr11\t63098519\t63138517\t-\tPLA2G16\t11145\tLung cancer\n+chr11\t63098519\t63138485\t-\tPLA2G16\t11145\tLung cancer\n+chr4\t124033248\t124063209\t-\tNUDT6\t11162\tLung cancer\n+chr4\t124033248\t124063573\t-\tNUDT6\t11162\tLung cancer\n+chr12\t26096757\t26117073\t+\tRASSF8\t11228\tLung cancer\n+chr12\t26017954\t26117073\t+\tRASSF8\t11228\tLung cancer\n+chr12\t26003230\t26117073\t+\tRASSF8\t11228\tLung cancer\n+chr12\t26003230\t26124091\t+\tRASSF8\t11228\tLung cancer\n+chr3\t50337344\t50340672\t-\tTUSC2\t11334\tLung cancer\n+chr15\t76674701\t76700692\t-\tCHRNA3\t1136\tLung cancer\n+chr15\t76672451\t76700692\t-\tCHRNA3\t1136\tLung cancer\n+chr15\t76644960\t76673515\t+\tCHRNA5\t1138\tLung cancer\n+chr17\t7289129\t7301656\t+\tCHRNB1\t1140\tLung cancer\n+chr10\t101938113\t101979334\t-\tCHUK\t1147\tInfectious lung disease\n+chr18\t894943\t902173\t+\tADCYAP1\t116\tLung cancer\n+chr18\t895296\t902173\t+\tADCYAP1\t116\tLung cancer\n+chr3\t46386636\t46392701\t+\tCCR5\t1234\tLung cancer\n+chr3_random\t664914\t670974\t+\tCCR5\t1234\tLung cancer\n+chr3\t46386636\t46392701\t+\tCCR5\t1234\tLung cancer\n+chr3_random\t664914\t670974\t+\tCCR5\t1234\tLung cancer\n+chr4\t100211152\t100228954\t-\tADH5\t128\tLung cancer\n+chr16\t3715056\t3870122\t-\tCREBBP\t1387\tLung cancer\n+chr16\t3715056\t3870122\t-\tCREBBP\t1387\tLung cancer\n+chr4\t5873391\t5945686\t-\tCRMP1\t1400\tLung cancer\n+chr4\t5873391\t5941216\t-\tCRMP1\t1400\tLung cancer\n+chr5\t131437383\t131439758\t+\tCSF2\t1437\tLung cancer\n+chr17\t35425213\t35427592\t+\tCSF3\t1440\tLung cancer\n+chr17\t35425213\t35427592\t+\tCSF3\t1440\tLung cancer\n+chr17\t35425213\t35427592\t+\tCSF3\t1440\tLung cancer\n+chr3\t123526700\t123543505\t+\tCSTA\t1475\tLung cancer\n+chr21\t44018258\t44020687\t-\tCSTB\t1476\tLung cancer\n+chr6\t132311009\t132314211\t-\tCTGF\t1490\tLung cancer\n+chr5\t148186348\t148188389\t+\tADRB2\t154\tLung cancer\n+chr16\t49333461\t49393347\t+\tCYLD\t1540\tInfectious lung disease\n+chr16\t49333529\t49393347\t+\tCYLD\t1540\tInfectious lung disease\n+chr16\t49333461\t49393347\t+\tCYLD\t1540\tInfectious lung disease\n+chr19\t46041282\t46048192\t-\tCYP2A6\t1548\tLung cancer\n+chr19\t46286207\t46293939\t+\tCYP2A13\t1553\tLung cancer\n+chr12\t56442383\t56447243\t-\tCYP27B1\t1594\tLung cancer\n+chr17\t58908165\t58928711\t+\tACE\t1636\tInfectious lung disease\n+chr17\t58915908\t58928711\t+\tACE\t1636\tInfectious lung disease\n+chr8\t6860804\t6863226\t-\tDEFA3\t1668\tLung disease\n+chr6\t32256723\t32260001\t-\tAGER\t177\tLung cancer\n+chr6_cox_hap1\t3555933\t3559211\t-\tAGER\t177\tLung cancer\n+chr6_qbl_hap2\t3363712\t3366990\t-\tAGER\t177\tLung cancer\n+chr6\t32256723\t32260001\t-\tAGER\t177\tLung cancer\n+chr6_cox_hap1\t3555933\t3559211\t-\tAGER\t177\tLung cancer\n+chr6_qbl_hap2\t3363712\t3366990\t-\tAGER\t177\tLung cancer\n+chr6\t7486868\t7531945\t+\tDSP\t1832\tLung cancer\n+chr6\t7486868\t7531945\t+\tDSP\t1832\tLung cancer\n+chr13\t77367616\t77447665\t-\tEDNRB\t1910\tLung cancer\n+chr13\t77367616\t77390967\t-\tEDNRB\t1910\tLung cancer\n+chr13\t77367616\t77390967\t-\t'..b'1189\t+\tC2\t717\tInfectious lung disease\n+chr6\t32003244\t32021427\t+\tC2\t717\tInfectious lung disease\n+chr6_cox_hap1\t3341528\t3359711\t+\tC2\t717\tInfectious lung disease\n+chr6_qbl_hap2\t3143005\t3161189\t+\tC2\t717\tInfectious lung disease\n+chr18\t27425727\t27432982\t+\tTTR\t7276\tLung cancer\n+chr11\t1200871\t1239982\t+\tMUC5B\t727897\tLung disease\n+chr6\t3098900\t3102782\t-\tTUBB2A\t7280\tLung cancer\n+chr10\t80985613\t80990169\t-\tSFTPA2\t729238\tLung disease\n+chr10\t81305559\t81310114\t-\tSFTPA2\t729238\tLung disease\n+chr9\t112046130\t112058599\t-\tTXN\t7295\tInfectious lung disease\n+chr9\t112046130\t112058599\t-\tTXN\t7295\tLung cancer\n+chr4\t40953654\t40965202\t+\tUCHL1\t7345\tLung cancer\n+chr9\t99477011\t99499512\t-\tXPA\t7507\tLung cancer\n+chr9\t99477012\t99499512\t-\tXPA\t7507\tLung cancer\n+chr6\t30959839\t30975912\t+\tDDR1\t780\tLung disease\n+chr6_cox_hap1\t2300465\t2316540\t+\tDDR1\t780\tLung disease\n+chr6_qbl_hap2\t2098794\t2114869\t+\tDDR1\t780\tLung disease\n+chr6\t30960305\t30975912\t+\tDDR1\t780\tLung disease\n+chr6_cox_hap1\t2300931\t2316540\t+\tDDR1\t780\tLung disease\n+chr6_qbl_hap2\t2099260\t2114869\t+\tDDR1\t780\tLung disease\n+chr6\t30964443\t30975912\t+\tDDR1\t780\tLung disease\n+chr6_cox_hap1\t2305069\t2316540\t+\tDDR1\t780\tLung disease\n+chr6_qbl_hap2\t2103398\t2114869\t+\tDDR1\t780\tLung disease\n+chr12\t26164227\t26169270\t-\tBHLHE41\t79365\tLung cancer\n+chr11\t14946625\t14950408\t-\tCALCA\t796\tInfectious lung disease\n+chr11\t14946625\t14950408\t-\tCALCA\t796\tLung cancer\n+chr11\t14944791\t14950408\t-\tCALCA\t796\tInfectious lung disease\n+chr11\t14944791\t14950408\t-\tCALCA\t796\tLung cancer\n+chr11\t14946623\t14950408\t-\tCALCA\t796\tInfectious lung disease\n+chr11\t14946623\t14950408\t-\tCALCA\t796\tLung cancer\n+chr10\t14960904\t14986310\t+\tSUV39H2\t79723\tLung cancer\n+chr17\t43403427\t43414146\t+\tCDK5RAP3\t80279\tLung cancer\n+chr11\t104370176\t104399105\t-\tCASP5\t838\tLung cancer\n+chr11\t104370176\t104399105\t-\tCASP5\t838\tLung cancer\n+chr11\t104370176\t104399105\t-\tCASP5\t838\tLung cancer\n+chr11\t104370176\t104399105\t-\tCASP5\t838\tLung cancer\n+chr11\t104370176\t104399105\t-\tCASP5\t838\tLung cancer\n+chr11\t104370176\t104399105\t-\tCASP5\t838\tLung cancer\n+chr1\t15691381\t15723377\t-\tCASP9\t842\tLung cancer\n+chr1\t15691381\t15723377\t-\tCASP9\t842\tLung cancer\n+chr21\t43346371\t43369109\t-\tCBS\t875\tLung cancer\n+chr10\t103103814\t103307060\t+\tBTRC\t8945\tLung cancer\n+chr10\t103103814\t103307060\t+\tBTRC\t8945\tLung cancer\n+chr5\t86725837\t86744592\t-\tCCNH\t902\tLung cancer\n+chr17\t73864456\t73867753\t-\tSOCS3\t9021\tLung cancer\n+chr11\t1367704\t1439904\t+\tBRSK2\t9024\tLung cancer\n+chr12\t47373018\t47397048\t-\tCCNT1\t904\tLung disease\n+chrX\t129091019\t129099696\t-\tAIFM1\t9131\tLung cancer\n+chrX\t129091019\t129127489\t-\tAIFM1\t9131\tLung cancer\n+chrX\t129091018\t129127489\t-\tAIFM1\t9131\tLung cancer\n+chrX\t129091018\t129127489\t-\tAIFM1\t9131\tLung cancer\n+chrX\t129091018\t129127489\t-\tAIFM1\t9131\tLung cancer\n+chr15\t57184611\t57204536\t+\tCCNB2\t9133\tLung cancer\n+chr11\t117680504\t117692100\t+\tCD3E\t916\tLung cancer\n+chr3\t50375234\t50515896\t-\tCACNA2D2\t9254\tLung cancer\n+chr3\t50375235\t50515896\t-\tCACNA2D2\t9254\tLung cancer\n+chr3\t50375234\t50515896\t-\tCACNA2D2\t9254\tLung cancer\n+chr6\t14225843\t14245127\t+\tCD83\t9308\tLung cancer\n+chr6\t14225843\t14245127\t+\tCD83\t9308\tLung cancer\n+chr3\t188043156\t188058946\t+\tADIPOQ\t9370\tLung cancer\n+chr1\t87100715\t87152695\t-\tSEP15\t9403\tLung cancer\n+chr1\t87100715\t87152695\t-\tSEP15\t9403\tLung cancer\n+chr19\t55553545\t55560743\t-\tNAPSA\t9476\tLung cancer\n+chr12\t51948349\t51973694\t+\tESPL1\t9700\tLung cancer\n+chr20\t4708669\t4752291\t-\tRASSF2\t9770\tLung cancer\n+chr20\t4708668\t4743769\t-\tRASSF2\t9770\tLung cancer\n+chr19\t10457795\t10474481\t-\tKEAP1\t9817\tLung cancer\n+chr19\t10457795\t10475054\t-\tKEAP1\t9817\tLung cancer\n+chr1\t155171255\t155281786\t-\tARHGEF11\t9826\tLung cancer\n+chr1\t155171255\t155281786\t-\tARHGEF11\t9826\tLung cancer\n+chr3\t38055699\t38139232\t+\tDLEC1\t9940\tLung cancer\n+chr3\t38055699\t38139232\t+\tDLEC1\t9940\tLung cancer\n+chrX\t140819307\t140824851\t+\tMAGEC1\t9947\tLung cancer\n+chr7\t86619612\t86663582\t+\tDMTF1\t9988\tLung cancer\n+chr7\t86619612\t86663582\t+\tDMTF1\t9988\tLung cancer\n+chr7\t86619805\t86663582\t+\tDMTF1\t9988\tLung cancer\n+chr7\t86619612\t86663582\t+\tDMTF1\t9988\tLung cancer\n+chr7\t86619612\t86663582\t+\tDMTF1\t9988\tLung cancer\n' |
| b |
| diff -r 000000000000 -r 64133669255f tool-data/funDo.loc.sample --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tool-data/funDo.loc.sample Wed Sep 25 11:22:47 2013 -0400 |
| b |
| @@ -0,0 +1,11 @@ +#This is a sample file distributed with Galaxy that is used by the FunDO tool. +#The funDo.loc file has this format (white space characters are TAB +#characters): +# +#<build> <description> <path to disease associated genes file> +# +#Your funDo.loc file should include an entry per line for each disease +#associated genes file you have stored. +# +#hg18 disease associated genes /galaxy/data/hg18/misc/funDo/genes-disease.Sept2010.interval +#hg19 disease associated genes /galaxy/data/hg19/misc/funDo/genes-disease.Sept2010.interval |