| Previous changeset 0:196795831b6a (2024-02-14) |
|
Commit message:
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/tapscan commit af8605d266717ed3453bd2a0947fed79c7098fb3 |
|
modified:
tapscan.xml test-data/output.2.tsv test-data/output.domtbl.tsv |
|
added:
tapscan_classify.pl tapscan_coverage_values_v11.txt tapscan_domains_v13.txt.gz tapscan_rules_v82.txt |
|
removed:
tapscan_coverage_values_v10.txt tapscan_domains_v12.txt.gz tapscan_rules_v81.txt tapscan_script_v74.pl |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan.xml --- a/tapscan.xml Wed Feb 14 13:54:16 2024 +0000 +++ b/tapscan.xml Thu Feb 22 10:07:53 2024 +0000 |
| [ |
| @@ -1,4 +1,4 @@ -<tool id="tapscan_classify" name="TAPScan Classify" version="4.74+galaxy0" profile="23.0"> +<tool id="tapscan_classify" name="TAPScan Classify" version="4.76+galaxy0" profile="23.0"> <description>Detect Transcription Associated Proteins (TAPs)</description> <edam_topics> <edam_topic>topic_0121</edam_topic> @@ -9,28 +9,28 @@ <requirement type="package" version="4.8">sed</requirement> </requirements> <required_files> - <include type="literal" path="tapscan_script_v74.pl"/> - <include type="literal" path="tapscan_domains_v12.txt"/> - <include type="literal" path="tapscan_rules_v81.txt"/> - <include type="literal" path="tapscan_coverage_values_v10.txt"/> + <include type="literal" path="tapscan_classify.pl"/> + <include type="literal" path="tapscan_domains_v13.txt.gz"/> + <include type="literal" path="tapscan_rules_v82.txt"/> + <include type="literal" path="tapscan_coverage_values_v11.txt"/> </required_files> <command detect_errors="aggressive"><![CDATA[ hmmsearch --domtblout domtblout.txt --cut_ga - '${__tool_directory__}/tapscan_domains_v12.txt.gz' + '${__tool_directory__}/tapscan_domains_v13.txt.gz' '$protein_fasta_in' && -perl '${__tool_directory__}/tapscan_script_v74.pl' +perl '${__tool_directory__}/tapscan_classify.pl' domtblout.txt - '${__tool_directory__}/tapscan_rules_v81.txt' + '${__tool_directory__}/tapscan_rules_v82.txt' '$taps_detected' '$taps_family_counts' '$taps_detected_extra' - '${__tool_directory__}/tapscan_coverage_values_v10.txt' + '${__tool_directory__}/tapscan_coverage_values_v11.txt' && |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_classify.pl --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tapscan_classify.pl Thu Feb 22 10:07:53 2024 +0000 |
| [ |
| b'@@ -0,0 +1,932 @@\n+#!/usr/bin/perl\n+use strict;\n+use warnings;\n+use File::Basename;\n+\n+my $tapscan_version = "v4.76";\n+print "Running TAPscan Classify version $tapscan_version \\n\\n";\n+\n+\n+# Written by Gerrit Timmerhaus (gerrit.timmerhaus@biologie.uni-freiburg.de).\n+# Changes included by Kristian Ullrich, Per Wilhelmsson, Romy Petroll and Saskia Hiltemann.\n+\n+# Script to extract all detected domains out of a hmmsearch results file and classify the families of all used proteins based on these domains.\n+# The classification depends on a table which contains all known classification rules for the protein families of interest and on specific coverage values defined for every domain.\n+# The script provides three outputs, namely output.1, output.2 and output.3. The output files are tables in ";"-delimited format.\n+# The structure of output.1 is: "sequence ID ; TAP family ; number of classifications ; domains".\n+# Output.3 shares in principle the same structure as output.1, except that subfamilies are considered. ("sequence ID ; TAP family ; Subfamily ; number of classifications ; domains")\n+# The superior TAP family is specified first, followed by the subfamily. If a TAP family has no subfamily, the TAP family is specified first and then a "-".\n+# The structure of output.2 is: "TAP family";"number of detected proteins".\n+# More than one entry for a protein is possible because the classification rules may allow more than one classification.\n+#\n+# The script must be startet with the arguments <hmmsearch output file> <classification rules> <output classifications file> <output family statistics file> <output subfamily classifications file> <"filter" if desired>\n+\n+if (!@ARGV or ($ARGV [0] eq "-h") or ($ARGV [0] eq "-help")) {\n+\tprint "Usage: extract.and.classify.pl <hmmsearch output file> <classification rules> <output classifications file> <output family statistics file> <output subfamily classifications file> <\\"filter\\" (if desired)>\\n\\n";\n+\texit;\n+}\n+\n+# hmmsearch_output: domtblout file\n+my $hmmsearch_output = $ARGV [0];\n+# decision_table: rules file\n+my $decision_table = $ARGV [1];\n+# family_classifications: output.1\n+my $family_classifications = $ARGV [2];\n+# family_statistics: output.2\n+my $family_statistics = $ARGV [3];\n+# subfamily_classifications: output.3\n+my $subfamily_classifications = $ARGV [4];\n+# domspec_cuts: coverage values file\n+my $domspec_cuts = $ARGV [5];\n+# gene_model_filte: filter for ARATH and ORYSA\n+my $gene_model_filter = $ARGV [6];\n+\n+# get basename for output files\n+my($basename, $dirs, $suffix) = fileparse($hmmsearch_output, qr/\\.[^.]*/);\n+\n+if ($family_statistics eq "") {\n+\tprint "Usage: extract.and.classify.pl <hmmsearch output file> <classification rules> <output classifications file> <output family statistics file> <output subfamil classifications file> <\\"filter\\" (if desired)>\\n\\n";\n+\texit;\n+}\n+\n+if ($gene_model_filter and $gene_model_filter eq "filter") {\n+\tprint "\\nGene model filter is activated. It only works for TAIR (Arabidopsis) and TIGR (Rice) proteins up to now\\n";\n+}\n+\n+# Array where the $hmmsearch-output/domtblout file will be stored\n+my @output = ();\n+# Array with domain-specific coverage values\n+my @cuts = ();\n+# Array with rules\n+my @dec_table = ();\n+# Counter for the number of detected domains in the hmmsearch output file\n+my $entry_counter = 0;\n+# Containes the actual result for a query sequence\n+my $akt_entry = "";\n+# Used to define query entry to ignore similar domains\n+my $whole_entry = "";\n+# Includes the final entries after ignoring similar domains\n+my @results_of_extraction = ();\n+# Used to define query entry to ignore similar domains\n+my $extracted_domain = "";\n+# Used to define query entry to ignore similar domains\n+my $present = "";\n+# Used to define query entry to ignore similar domains\n+my $protein = "";\n+\n+my $lek = "";\n+\n+############################################\n+### 1. Read in the hmmsearch output file ###\n+############################################\n+\n+print "\\n*** '..b'milies into $family_list to create output.2\n+foreach my $fcf_line (@family_classifications_file) {\n+\t$fcf_line =~ /^[^;]+;([^;]+)/;\n+\t#push @family_list, "$1";\n+\tprint FAMILY_CLASSIFICATIONS "$fcf_line";\n+}\n+close (FAMILY_CLASSIFICATIONS);\n+\n+foreach my $fcf_line (@subfamily_classifications_file) {\n+\t$fcf_line =~ /^[^;]+;([^;]+);([^;]+)/;\n+\tif ($2 eq "-") {\n+\t\tpush @family_list, "$1";\n+\t}\n+\telse {\n+\t\tpush @family_list, "$2";\n+\t}\n+\tprint SUBFAMILY_CLASSIFICATIONS "$fcf_line";\n+}\n+\n+close (SUBFAMILY_CLASSIFICATIONS);\n+\n+print "*** calculating the family statistics and write it in $family_statistics ***\\n\\n";\n+\n+##################################\n+### 5. Create the output files ###\n+##################################\n+\n+my $statistics_outputfile = "$family_statistics";\n+\n+unless (open(FAMILY_STATISTICS, ">$statistics_outputfile")) {\n+\tprint "Cannot open file \\"$statistics_outputfile\\" to write to!!\\n\\n";\n+\texit;\n+}\n+\n+# Count the family entries\n+my @output_family_statistics = ();\n+my @gefundene_familien = ();\n+my $family_counter = 1;\n+\n+shift @family_list;\n+@family_list = sort @family_list;\n+\n+\n+my $old_family = "";\n+push @family_list, \'BAD FIX\'; # Makes to loop go through every fam and stops at non fam(BAD FIX).\n+\n+foreach my $line (@family_list) {\n+\tif ($line eq $old_family) {\n+\t\t$family_counter++;\n+\t}\n+\telsif ($old_family ne "") {\n+\t\tpush (@output_family_statistics,"$old_family;$family_counter\\n");\n+\t\t# Add all found families to a list of found families\n+\t\tpush (@gefundene_familien,"$old_family");\n+\t $family_counter=1;\n+\t}\n+\t$old_family = $line;\n+}\n+\n+my %hash = ();\n+\n+# Put all families from the classifictaion ruled and all found families in a hash\n+foreach my $element (@gefundene_familien,@liste_alle_familien) {$hash{$element}++;}\n+\n+# Remove merged families\n+delete $hash{\'GARP_ARR-B_Myb\'};\n+delete $hash{\'GARP_ARR-B_G2\'};\n+delete $hash{\'bZIP1\'};\n+delete $hash{\'bZIP2\'};\n+delete $hash{\'bZIPAUREO\'};\n+delete $hash{\'bZIPCDD\'};\n+delete $hash{\'HRT\'};\n+delete $hash{\'GIY_YIG\'};\n+\n+# Add all not found families from the classification rules to @output_family_statistics\n+# With zero as number of families found\n+\n+foreach my $element (keys %hash) {\n+\tif (($hash{$element} == 1) and ( $element eq "0_no_family_found")) {\n+\tpush (@output_family_statistics,"$element;$unclassified_families\\n");\n+\t}\n+\tif (($hash{$element} == 1) and ($element ne "0_no_family_found")) {\n+\tpush (@output_family_statistics,"$element;0\\n");\n+\t}\n+}\n+\n+# Sort @output_family_statistics caseinsensitive-alphabetically\n+my @sortierte_statistik = sort {lc $a cmp lc $b} @output_family_statistics;\n+\n+# Print headline to @sortierte_statistik\n+unshift (@sortierte_statistik,"family statistics for $hmmsearch_output\\n");\n+\n+print FAMILY_STATISTICS @sortierte_statistik;\n+\n+# Print FAMILY_STATISTICS "$old_family;$family_counter\\n";\n+\n+close (FAMILY_STATISTICS);\n+\n+#################################################\n+### 6. Give out some statistical informations ###\n+#################################################\n+\n+$entry_counter [0] = 0;\n+my $sum = 0;\n+foreach my $entry (@entry_counter) {\n+\t$sum += $entry;\n+}\n+print "$classified_families classifications were found for $sum proteins.\\n";\n+print "This classifications are divided in:\\n";\n+my $count = 0;\n+foreach my $element (@entry_counter) {\n+\tif ($count != 0) {\n+\t\tprint "$element proteins were classified for $count";\n+\t\tif ($count == 1) {print " family\\n";}\n+\t\telse {print " different families\\n";}\n+\t}\n+\t$count++;\n+}\n+print "\\n$unclassified_families proteins could not be classified\\n\\n";\n+\n+print "*** The results were written in $family_classifications and $subfamily_classifications ***\\n";\n+print "*** done ***\\n\\n";\n+\n+\n+exit;\n+\n+sub get_file_data {\n+\n+\tmy ($filename) = @_;\n+\n+\tuse strict;\n+\tuse warnings;\n+\n+\tmy @filedata = ();\n+\n+\tunless( open(GET_FILE_DATA, $filename)) {\n+\t\tprint STDERR "Cannot open file \\"$filename\\"n\\n";\n+\t\texit;\n+\t}\n+\n+\t@filedata = <GET_FILE_DATA>;\n+\n+\tclose GET_FILE_DATA;\n+\n+\treturn @filedata;\n+}\n+\n+\n' |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_coverage_values_v10.txt --- a/tapscan_coverage_values_v10.txt Wed Feb 14 13:54:16 2024 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 |
| b |
| @@ -1,154 +0,0 @@ -Acetyltransf_1 0.0931034483 -AP2 0.3214285714 -ARID 0.0516393443 -AUX_IAA 0.1077981651 -Auxin_resp 0.4308510638 -B3 0.196875 -BES1_N 0.3888888889 -BSD 0.485915493 -BTB 0.3105095541 -bZIP_1 0.64453125 -bZIP_2 0.5948275862 -C1_2 0.1612903226 -CAF1C_H4-bd 0.1764705882 -CBFB_NFYA 0.2019230769 -CCT 0.4952830189 -CG-1 0.5568181818 -CSD 0.5878378378 -DDT 0.2678571429 -DEAD 0.1115591398 -dsrm 0.2330097087 -DUF260 0.3638059701 -DUF296 0.155075188 -DUF547 0.1181672026 -DUF573 0.3515625 -DUF632 0.3302603037 -DUF702 0.2314126394 -E2F_TDP 0.2078313253 -EIN3 0.5874125874 -FHA 0.1777456647 -FLO_LFY 0.4569138277 -FYRC 0.3928571429 -FYRN 0.46875 -GAGA_bind 0.2265774379 -GATA 0.5406976744 -GRAS 0.3990825688 -Helicase_C 0.118852459 -HLH 0.3191489362 -HMG_box 0.6866197183 -Homeobox 0.5485074627 -HSF_DNA-bind 0.071641791 -IQ 0.6428571429 -JmjC 0.5058139535 -JmjN 0.4261363636 -K-box 0.438 -KNOX1 0.578125 -KNOX2 0.6346153846 -LIM 0.4726027397 -MBF1 0.3049450549 -Med26 0.1939655172 -Med31 0.451048951 -Med6 0.1577868852 -Med7 0.1818181818 -MEKHLA 0.4357541899 -mTERF 0.4850543478 -Myb_DNA-binding 0.3636363636 -NAM 0.0142857143 -O-FucT 0.0914866582 -Ovate 0.3137755102 -PAH 0.1194267516 -PAZ 0.1564569536 -PC4 0.3848684211 -PHD 0.37 -Piwi 0.4082278481 -PLATZ 0.30625 -PP2C 0.4110962567 -QLQ 0.61875 -RB_B 0.1663987138 -Rcd1 0.4131355932 -Response_reg 0.5017241379 -RFX_DNA_binding 0.5257731959 -RHD_DNA_bind 0.4581447964 -Ribonuclease_3 0.0935013263 -RRN3 0.2342427093 -Runt 0.5714285714 -RWP-RK 0.4542253521 -S1FA 0.7386363636 -SBP 0.4274193548 -SET 0.0392857143 -SH2 0.4197247706 -Sigma70_r2 0.1675531915 -Sigma70_r3 0.6346153846 -Sigma70_r4 0.6388888889 -SIR2 0.45703125 -SNF2_N 0.1150895141 -SRF-TF 0.406779661 -SSXT 0.5856164384 -START 0.5067567568 -STAT_bind 0.3781869688 -SWIB 0.3267326733 -SWIRM 0.2532467532 -TANGO2 0.125 -TCP 0.219665272 -TCR 0.4943181818 -TEA 0.4078282828 -TF_AP-2 0.5660377358 -Tfb2 0.1757668712 -tify 0.5955882353 -Tub 0.0872576177 -VEFS-Box 0.4918831169 -WD40 0.0563909774 -WHIM1 0.5263157895 -Whirly 0.5648148148 -WRC 0.6223404255 -WRKY 0.2196969697 -WSD 0.1768867925 -YABBY 0.2804621849 -zf-AN1 0.4027777778 -zf-B_box 0.2746478873 -zf-C2H2 0.5080645161 -zf-C5HC2 0.253125 -zf-CCCH 0.5689655172 -zf-Dof 0.5955882353 -ZF-HD_dimer 0.5294117647 -zf-MIZ 0.6805555556 -zf-TAZ 0.1566455696 -zf-ZPR1 0.1608910891 -Zn_clus 0.5480769231 -Alfin-like 0.75 -BEL 0.75 -DNC 0.75 -FIE_clipped_for_HMM 0.75 -G2-like_Domain 0.75 -HRT 0.75 -KNOXC 0.75 -LUFS_Domain 0.75 -NF-YB 0.75 -NF-YC 0.75 -NOZZLE 0.75 -PINTOX 0.75 -STER_AP 0.75 -trihelix 0.75 -ULT_Domain 0.75 -VARL 0.75 -VOZ_Domain 0.75 -WOX_HD 0.75 -CXC 0.75 -bZIP_AUREO 0.75 -bZIP_CDD 0.50 -ALOG 0.75 -C2H2-IDD 0.75 -zf-MYST 0.75 -CBP 0.75 -DUF3591 0.75 -LOB2 0.75 -zz-ADA2 0.75 -NLP 0.75 -CRF 0.75 -GIY_YIG 0.75 -ZPR 0.75 -LD 0.75 -NDX 0.75 -SAWADEE 0.75 -C1HDZ 0.75 -C2HDZ 0.75 |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_coverage_values_v11.txt --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tapscan_coverage_values_v11.txt Thu Feb 22 10:07:53 2024 +0000 |
| b |
| @@ -0,0 +1,155 @@ +Acetyltransf_1 0.0931034483 +AP2 0.3214285714 +ARID 0.0516393443 +AUX_IAA 0.1077981651 +Auxin_resp 0.4308510638 +B3 0.196875 +BES1_N 0.3888888889 +BSD 0.485915493 +BTB 0.3105095541 +bZIP_1 0.64453125 +bZIP_2 0.5948275862 +C1_2 0.1612903226 +CAF1C_H4-bd 0.1764705882 +CBFB_NFYA 0.2019230769 +CCT 0.4952830189 +CG-1 0.5568181818 +CSD 0.5878378378 +DDT 0.2678571429 +DEAD 0.1115591398 +dsrm 0.2330097087 +DUF260 0.3638059701 +DUF296 0.155075188 +DUF547 0.1181672026 +DUF573 0.3515625 +DUF632 0.3302603037 +DUF702 0.2314126394 +E2F_TDP 0.2078313253 +EIN3 0.5874125874 +FHA 0.1777456647 +FLO_LFY 0.4569138277 +FYRC 0.3928571429 +FYRN 0.46875 +GAGA_bind 0.2265774379 +GATA 0.5406976744 +GRAS 0.3990825688 +Helicase_C 0.118852459 +HLH 0.3191489362 +HMG_box 0.6866197183 +Homeobox 0.5485074627 +HSF_DNA-bind 0.071641791 +IQ 0.6428571429 +JmjC 0.5058139535 +JmjN 0.4261363636 +K-box 0.438 +KNOX1 0.578125 +KNOX2 0.6346153846 +LIM 0.4726027397 +MBF1 0.3049450549 +Med26 0.1939655172 +Med31 0.451048951 +Med6 0.1577868852 +Med7 0.1818181818 +MEKHLA 0.4357541899 +mTERF 0.4850543478 +Myb_DNA-binding 0.3636363636 +NAM 0.0142857143 +O-FucT 0.0914866582 +Ovate 0.3137755102 +PAH 0.1194267516 +PAZ 0.1564569536 +PC4 0.3848684211 +PHD 0.37 +Piwi 0.4082278481 +PLATZ 0.30625 +PP2C 0.4110962567 +QLQ 0.61875 +RB_B 0.1663987138 +Rcd1 0.4131355932 +Response_reg 0.5017241379 +RFX_DNA_binding 0.5257731959 +RHD_DNA_bind 0.4581447964 +Ribonuclease_3 0.0935013263 +RRN3 0.2342427093 +Runt 0.5714285714 +RWP-RK 0.4542253521 +S1FA 0.7386363636 +SBP 0.4274193548 +SET 0.0392857143 +SH2 0.4197247706 +Sigma70_r2 0.1675531915 +Sigma70_r3 0.6346153846 +Sigma70_r4 0.6388888889 +SIR2 0.45703125 +SNF2_N 0.1150895141 +SRF-TF 0.406779661 +SSXT 0.5856164384 +START 0.5067567568 +STAT_bind 0.3781869688 +SWIB 0.3267326733 +SWIRM 0.2532467532 +TANGO2 0.125 +TCP 0.219665272 +TCR 0.4943181818 +TEA 0.4078282828 +TF_AP-2 0.5660377358 +Tfb2 0.1757668712 +tify 0.5955882353 +Tub 0.0872576177 +VEFS-Box 0.4918831169 +WD40 0.0563909774 +WHIM1 0.5263157895 +Whirly 0.5648148148 +WRC 0.6223404255 +WRKY 0.2196969697 +WSD 0.1768867925 +YABBY 0.2804621849 +zf-AN1 0.4027777778 +zf-B_box 0.2746478873 +zf-C2H2 0.5080645161 +zf-C5HC2 0.253125 +zf-CCCH 0.5689655172 +zf-Dof 0.5955882353 +ZF-HD_dimer 0.5294117647 +zf-MIZ 0.6805555556 +zf-TAZ 0.1566455696 +zf-ZPR1 0.1608910891 +Zn_clus 0.5480769231 +Alfin-like 0.75 +BEL 0.75 +DNC 0.75 +FIE_clipped_for_HMM 0.75 +G2-like_Domain 0.75 +HRT 0.75 +KNOXC 0.75 +LUFS_Domain 0.75 +NF-YB 0.75 +NF-YC 0.75 +NOZZLE 0.75 +PINTOX 0.75 +STER_AP 0.75 +trihelix 0.75 +ULT_Domain 0.75 +VARL 0.75 +VOZ_Domain 0.75 +WOX_HD 0.75 +CXC 0.75 +bZIP_AUREO 0.75 +bZIP_CDD 0.50 +ALOG 0.75 +C2H2-IDD 0.75 +zf-MYST 0.75 +CBP 0.75 +DUF3591 0.75 +LOB2 0.75 +zz-ADA2 0.75 +NLP 0.75 +CRF 0.75 +GIY_YIG 0.75 +ZPR 0.75 +LD 0.75 +NDX 0.75 +SAWADEE 0.75 +C1HDZ 0.75 +C2HDZ 0.75 +Homeobox_KN 0.75 |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_domains_v12.txt.gz |
| b |
| Binary file tapscan_domains_v12.txt.gz has changed |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_domains_v13.txt.gz |
| b |
| Binary file tapscan_domains_v13.txt.gz has changed |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_rules_v81.txt --- a/tapscan_rules_v81.txt Wed Feb 14 13:54:16 2024 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 |
| b |
| @@ -1,298 +0,0 @@ -ABI3/VP1;AP2;should not -ABI3/VP1;Auxin_resp;should not -ABI3/VP1;B3;should -ABI3/VP1;WRKY;should not -Alfin-like;Alfin-like;should -Alfin-like;Homeobox;should not -Alfin-like;zf-TAZ;should not -Alfin-like;PHD;should not -AP2;AP2;should -AP2;CRF;should not -ARF;Auxin_resp;should -Argonaute;Piwi;should -Argonaute;PAZ;should -ARID;ARID;should -Aux/IAA;AUX_IAA;should -Aux/IAA;Auxin_resp;should not -Aux/IAA;B3;should not -BBR/BPC;GAGA_bind;should -BES1;BES1_N;should -bHLH;HLH;should -bHLH;TCP;should not -bHLH_TCP;TCP;should -bHSH;TF_AP-2;should -BSD domain containing;BSD;should -bZIP1;bZIP_1;should -bZIP1;HLH;should not -bZIP1;Homeobox;should not -bZIP2;bZIP_2;should -bZIP2;HLH;should not -bZIP2;Homeobox;should not -bZIPAUREO;bZIP_AUREO;should -bZIPAUREO;HLH;should not -bZIPAUREO;Homeobox;should not -bZIPCDD;bZIP_CDD;should -bZIPCDD;HLH;should not -bZIPCDD;Homeobox;should not -C2C2_CO-like;CCT;should -C2C2_CO-like;GATA;should not -C2C2_CO-like;tify;should not -C2C2_CO-like;PLATZ;should not -C2C2_CO-like;zf-B_box;should -C2C2_Dof;zf-Dof;should -C2C2_Dof;GATA;should not -C2C2_GATA;GATA;should -C2C2_GATA;tify;should not -C2C2_GATA;zf-Dof;should not -C2C2_YABBY;YABBY;should -C2H2;zf-C2H2;should -C2H2;zf-MIZ;should not -C3H;AP2;should not -C3H;SRF-TF;should not -C3H;MYB-2R;should not -C3H;MYB-3R;should not -C3H;MYB-4R;should not -C3H;zf-C2H2;should not -C3H;zf-CCCH;should -CAMTA;CG-1;should -CAMTA;IQ;should -NF-YA;bZIP_1;should not -NF-YA;bZIP_2;should not -NF-YA;CBFB_NFYA;should -NF-YB;NF-YB;should -NF-YB;NF-YC;should not -NF-YC;NF-YB;should not -NF-YC;NF-YC;should -NF-YC;HMG_box;should not -Coactivator p15;PC4;should -CPP;TCR;should -CSD;CSD;should -CudA;STAT_bind;should -CudA;SH2;should -DBP;DNC;should -DBP;PP2C;should -DDT;DDT;should -DDT;Homeobox;should not -DDT;Alfin-like;should not -Dicer;Piwi;should not -Dicer;DEAD;should -Dicer;Helicase_C;should -Dicer;Ribonuclease_3;should -Dicer;dsrm;should -DUF246 domain containing/O-FucT;O-FucT;should -DUF296 domain containing;DUF296;should -DUF547 domain containing;DUF547;should -DUF632 domain containing;DUF632;should -DUF833 domain containing/TANGO2;TANGO2;should -E2F/DP;E2F_TDP;should -EIL;EIN3;should -FHA;FHA;should -GARP_ARR-B_G2;CCT;should not -GARP_ARR-B_G2;G2-like_Domain;should -GARP_ARR-B_G2;Response_reg;should -GARP_ARR-B_Myb;CCT;should not -GARP_ARR-B_Myb;Response_reg;should -GARP_ARR-B_Myb;Myb_DNA-binding;should -GARP_G2-like;G2-like_Domain;should -GARP_G2-like;Response_reg;should not -GARP_G2-like;Myb_DNA-binding;should not -GeBP;DUF573;should -GIF;SSXT;should -GNAT;Acetyltransf_1;should -GNAT;PHD;should not -GRAS;GRAS;should -GRF;QLQ;should -GRF;WRC;should -C3HDZ;Homeobox;should -C3HDZ;START;should -C3HDZ;MEKHLA;should -C4HDZ;Homeobox;should -C4HDZ;START;should -C4HDZ;MEKHLA;should not -HD_PLINC;ZF-HD_dimer;should -HD_WOX;WOX_HD;should -HD_DDT;Homeobox;should -HD_DDT;DDT;should -HD_DDT;WHIM1;should -HD_DDT;WSD;should -HD_PHD;PHD;should -HD_PHD;Homeobox;should -HD_PINTOX;Homeobox;should -HD_PINTOX;PINTOX;should -HD_BEL;Homeobox;should -HD_BEL;BEL;should -HD_KNOX1;Homeobox;should -HD_KNOX1;KNOX1;should -HD_KNOX1;KNOX2;should -HD_KNOX1;KNOXC;should not -HD_KNOX2;Homeobox;should -HD_KNOX2;KNOX1;should -HD_KNOX2;KNOX2;should -HD_KNOX2;KNOXC;should -HD-other;EIN3;should not -HD-other;Homeobox;should -HD-other;bZIP_1;should not -HD-other;WOX_HD;should not -HD-other;PINTOX;should not -HD-other;PHD;should not -HD-other;BEL;should not -HMG;ARID;should not -HMG;HMG_box;should -HMG;YABBY;should not -HRT;HRT;should -HSF;HSF_DNA-bind;should -IWS1;Med26;should -Jumonji_PKDM7;JmjC;should -Jumonji_PKDM7;JmjN;should -Jumonji_PKDM7;zf-C5HC2;should -Jumonji_PKDM7;FYRN;should -Jumonji_PKDM7;FYRC;should -Jumonji_Other;JmjC;should -LFY;FLO_LFY;should -LIM;two_or_more_LIM;should -LUG;LUFS_Domain;should -MADS;SRF-TF;should -MADS;K-box;should not -MADS_MIKC;SRF-TF;should -MADS_MIKC;K-box;should -MBF1;MBF1;should -Med6;Med6;should -Med7;Med7;should -mTERF;mTERF;should -MYB-2R;G2-like_Domain;should not -MYB-2R;Response_reg;should not -MYB-2R;trihelix;should not -MYB-2R;MYB-2R;should -MYB-3R;G2-like_Domain;should not -MYB-3R;Response_reg;should not -MYB-3R;trihelix;should not -MYB-3R;MYB-3R;should -MYB-4R;G2-like_Domain;should not -MYB-4R;Response_reg;should not -MYB-4R;trihelix;should not -MYB-4R;MYB-4R;should -MYB-related;ARID;should not -MYB-related;G2-like_Domain;should not -MYB-related;Myb_DNA-binding;should -MYB-related;Response_reg;should not -MYB-related;trihelix;should not -MYB-related;MYB-2R;should not -MYB-related;MYB-3R;should not -MYB-related;MYB-4R;should not -NAC;NAM;should -NZZ;NOZZLE;should -OFP;Ovate;should -PcG_EZ;CXC;should -PcG_EZ;SET;should -PcG_FIE;FIE_clipped_for_HMM;should -PcG_FIE;WD40;should -PcG_VEFS;VEFS-Box;should -PcG_VEFS;zf-C2H2;should not -PcG_MSI;WD40;should -PcG_MSI;CAF1C_H4-bd;should -PcG_MSI;FIE_clipped_for_HMM;should not -PHD;Myb_DNA-binding;should not -PHD;Alfin-like;should not -PHD;ARID;should not -PHD;DDT;should not -PHD;Homeobox;should not -PHD;JmjC;should not -PHD;JmjN;should not -PHD;PHD;should -PHD;SWIB;should not -PHD;zf-TAZ;should not -PHD;zf-MIZ;should not -PHD;zf-CCCH;should not -PHD;HMG_box;should not -PLATZ;PLATZ;should -Pseudo ARR-B;CCT;should -Pseudo ARR-B;Response_reg;should -Pseudo ARR-B;tify;should not -RB;RB_B;should -Rcd1-like;Rcd1;should -Rel;RHD_DNA_bind;should -RF-X;RFX_DNA_binding;should -RRN3;RRN3;should -Runt;Runt;should -S1Fa-like;S1FA;should -SAP;STER_AP;should -SBP;SBP;should -SET;zf-C2H2;should not -SET;TCR;should not -SET;CXC;should not -SET;PHD;should not -SET;Myb_DNA-binding;should not -SET;SET;should -Sigma70-like;Sigma70_r2;should -Sigma70-like;Sigma70_r3;should -Sigma70-like;Sigma70_r4;should -Sin3;PAH;should -Sin3;WRKY;should not -Sir2;SIR2;should -SOH1;Med31;should -SRS;DUF702;should -SWI/SNF_BAF60b;SWIB;should -SWI/SNF_SNF2;AP2;should not -SWI/SNF_SNF2;PHD;should not -SWI/SNF_SNF2;SNF2_N;should -SWI/SNF_SNF2;zf-CCCH;should not -SWI/SNF_SNF2;Myb_DNA-binding;should not -SWI/SNF_SNF2;HMG_box;should not -SWI/SNF_SWI3;SWIRM;should -SWI/SNF_SWI3;Myb_DNA-binding;should -TEA;TEA;should -TFb2;Tfb2;should -tify;tify;should -TRAF;BTB;should -TRAF;zf-TAZ;should not -Trihelix;trihelix;should -TUB;Tub;should -ULT;ULT_Domain;should -VARL;VARL;should -VOZ;VOZ_Domain;should -Whirly;Whirly;should -WRKY;WRKY;should -Zinc finger, AN1 and A20 type;zf-AN1;should -Zinc finger, AN1 and A20 type;zf-C2H2;should not -Zinc finger, MIZ type;zf-MIZ;should -Zinc finger, MIZ type;zf-C2H2;should not -Zinc finger, ZPR1;zf-ZPR1;should -Zn_clus;Zn_clus;should -ALOG;ALOG;should -C2H2;C2H2-IDD;should not -C2H2_IDD;C2H2-IDD;should -C2H2_IDD;zf-C2H2;should -MYST;zf-MYST;should -CBP;CBP;should -CBP;zf-TAZ;should -CBP;BTB;should not -TAFII250;DUF3591;should -LOB1;bZIP_1;should not -LOB1;bZIP_2;should not -LOB1;DUF260;should -LOB1;HLH;should not -LOB1;Homeobox;should not -LOB2;LOB2;should -LDL/FLD;SWIRM;should -LDL/FLD;Myb_DNA-binding;should not -ADA2;zz-ADA2;should -ADA2;Myb_DNA-binding;should -RKD;RWP-RK;should -RKD;NLP;should not -NLP;RWP-RK;should -NLP;NLP;should -CRF;CRF;should -CRF;AP2;should -GIY_YIG;GIY_YIG;should -ZPR;ZPR;should -HD-LD;LD;should -HD-NDX;NDX;should -HD-SAWADEE;SAWADEE;should -C1HDZ;C1HDZ;should -C1HDZ;Homeobox;should -C1HDZ;START;should not -C1HDZ;MEKHLA;should not -C2HDZ;C2HDZ;should -C2HDZ;Homeobox;should -C2HDZ;START;should not -C2HDZ;MEKHLA;should not |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_rules_v82.txt --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tapscan_rules_v82.txt Thu Feb 22 10:07:53 2024 +0000 |
| b |
| @@ -0,0 +1,302 @@ +ABI3/VP1;AP2;should not +ABI3/VP1;Auxin_resp;should not +ABI3/VP1;B3;should +ABI3/VP1;WRKY;should not +Alfin-like;Alfin-like;should +Alfin-like;Homeobox;should not +Alfin-like;zf-TAZ;should not +Alfin-like;PHD;should not +AP2;AP2;should +AP2;CRF;should not +ARF;Auxin_resp;should +Argonaute;Piwi;should +Argonaute;PAZ;should +ARID;ARID;should +Aux/IAA;AUX_IAA;should +Aux/IAA;Auxin_resp;should not +Aux/IAA;B3;should not +BBR/BPC;GAGA_bind;should +BES1;BES1_N;should +bHLH;HLH;should +bHLH;TCP;should not +bHLH_TCP;TCP;should +bHSH;TF_AP-2;should +BSD domain containing;BSD;should +bZIP1;bZIP_1;should +bZIP1;HLH;should not +bZIP1;Homeobox;should not +bZIP2;bZIP_2;should +bZIP2;HLH;should not +bZIP2;Homeobox;should not +bZIPAUREO;bZIP_AUREO;should +bZIPAUREO;HLH;should not +bZIPAUREO;Homeobox;should not +bZIPCDD;bZIP_CDD;should +bZIPCDD;HLH;should not +bZIPCDD;Homeobox;should not +C2C2_CO-like;CCT;should +C2C2_CO-like;GATA;should not +C2C2_CO-like;tify;should not +C2C2_CO-like;PLATZ;should not +C2C2_CO-like;zf-B_box;should +C2C2_Dof;zf-Dof;should +C2C2_Dof;GATA;should not +C2C2_GATA;GATA;should +C2C2_GATA;tify;should not +C2C2_GATA;zf-Dof;should not +C2C2_YABBY;YABBY;should +C2H2;zf-C2H2;should +C2H2;zf-MIZ;should not +C3H;AP2;should not +C3H;SRF-TF;should not +C3H;MYB-2R;should not +C3H;MYB-3R;should not +C3H;MYB-4R;should not +C3H;zf-C2H2;should not +C3H;zf-CCCH;should +CAMTA;CG-1;should +CAMTA;IQ;should +NF-YA;bZIP_1;should not +NF-YA;bZIP_2;should not +NF-YA;CBFB_NFYA;should +NF-YB;NF-YB;should +NF-YB;NF-YC;should not +NF-YC;NF-YB;should not +NF-YC;NF-YC;should +NF-YC;HMG_box;should not +Coactivator p15;PC4;should +CPP;TCR;should +CSD;CSD;should +CudA;STAT_bind;should +CudA;SH2;should +DBP;DNC;should +DBP;PP2C;should +DDT;DDT;should +DDT;Homeobox;should not +DDT;Alfin-like;should not +Dicer;Piwi;should not +Dicer;DEAD;should +Dicer;Helicase_C;should +Dicer;Ribonuclease_3;should +Dicer;dsrm;should +DUF246 domain containing/O-FucT;O-FucT;should +DUF296 domain containing;DUF296;should +DUF547 domain containing;DUF547;should +DUF632 domain containing;DUF632;should +DUF833 domain containing/TANGO2;TANGO2;should +E2F/DP;E2F_TDP;should +EIL;EIN3;should +FHA;FHA;should +GARP_ARR-B_G2;CCT;should not +GARP_ARR-B_G2;G2-like_Domain;should +GARP_ARR-B_G2;Response_reg;should +GARP_ARR-B_Myb;CCT;should not +GARP_ARR-B_Myb;Response_reg;should +GARP_ARR-B_Myb;Myb_DNA-binding;should +GARP_G2-like;G2-like_Domain;should +GARP_G2-like;Response_reg;should not +GARP_G2-like;Myb_DNA-binding;should not +GeBP;DUF573;should +GIF;SSXT;should +GNAT;Acetyltransf_1;should +GNAT;PHD;should not +GRAS;GRAS;should +GRF;QLQ;should +GRF;WRC;should +C3HDZ;Homeobox;should +C3HDZ;START;should +C3HDZ;MEKHLA;should +C4HDZ;Homeobox;should +C4HDZ;START;should +C4HDZ;MEKHLA;should not +HD_PLINC;ZF-HD_dimer;should +HD_WOX;WOX_HD;should +HD_DDT;Homeobox;should +HD_DDT;DDT;should +HD_DDT;WHIM1;should +HD_DDT;WSD;should +HD_PHD;PHD;should +HD_PHD;Homeobox;should +HD_PINTOX;Homeobox;should +HD_PINTOX;PINTOX;should +HD_TALE_BEL;Homeobox_KN;should +HD_TALE_BEL;BEL;should +HD_TALE_KNOX1;Homeobox_KN;should +HD_TALE_KNOX1;KNOX1;should +HD_TALE_KNOX1;KNOX2;should +HD_TALE_KNOX1;KNOXC;should not +HD_TALE_KNOX2;Homeobox_KN;should +HD_TALE_KNOX2;KNOX1;should +HD_TALE_KNOX2;KNOX2;should +HD_TALE_KNOX2;KNOXC;should +HD-other;EIN3;should not +HD-other;Homeobox;should +HD-other;bZIP_1;should not +HD-other;WOX_HD;should not +HD-other;PINTOX;should not +HD-other;PHD;should not +HD-other;BEL;should not +HD_TALE;Homeobox_KN;should +HD_TALE;BEL;should not +HD_TALE;KNOX1;should not +HD_TALE;KNOX2;should not +HMG;ARID;should not +HMG;HMG_box;should +HMG;YABBY;should not +HRT;HRT;should +HSF;HSF_DNA-bind;should +IWS1;Med26;should +Jumonji_PKDM7;JmjC;should +Jumonji_PKDM7;JmjN;should +Jumonji_PKDM7;zf-C5HC2;should +Jumonji_PKDM7;FYRN;should +Jumonji_PKDM7;FYRC;should +Jumonji_Other;JmjC;should +LFY;FLO_LFY;should +LIM;two_or_more_LIM;should +LUG;LUFS_Domain;should +MADS;SRF-TF;should +MADS;K-box;should not +MADS_MIKC;SRF-TF;should +MADS_MIKC;K-box;should +MBF1;MBF1;should +Med6;Med6;should +Med7;Med7;should +mTERF;mTERF;should +MYB-2R;G2-like_Domain;should not +MYB-2R;Response_reg;should not +MYB-2R;trihelix;should not +MYB-2R;MYB-2R;should +MYB-3R;G2-like_Domain;should not +MYB-3R;Response_reg;should not +MYB-3R;trihelix;should not +MYB-3R;MYB-3R;should +MYB-4R;G2-like_Domain;should not +MYB-4R;Response_reg;should not +MYB-4R;trihelix;should not +MYB-4R;MYB-4R;should +MYB-related;ARID;should not +MYB-related;G2-like_Domain;should not +MYB-related;Myb_DNA-binding;should +MYB-related;Response_reg;should not +MYB-related;trihelix;should not +MYB-related;MYB-2R;should not +MYB-related;MYB-3R;should not +MYB-related;MYB-4R;should not +NAC;NAM;should +NZZ;NOZZLE;should +OFP;Ovate;should +PcG_EZ;CXC;should +PcG_EZ;SET;should +PcG_FIE;FIE_clipped_for_HMM;should +PcG_FIE;WD40;should +PcG_VEFS;VEFS-Box;should +PcG_VEFS;zf-C2H2;should not +PcG_MSI;WD40;should +PcG_MSI;CAF1C_H4-bd;should +PcG_MSI;FIE_clipped_for_HMM;should not +PHD;Myb_DNA-binding;should not +PHD;Alfin-like;should not +PHD;ARID;should not +PHD;DDT;should not +PHD;Homeobox;should not +PHD;JmjC;should not +PHD;JmjN;should not +PHD;PHD;should +PHD;SWIB;should not +PHD;zf-TAZ;should not +PHD;zf-MIZ;should not +PHD;zf-CCCH;should not +PHD;HMG_box;should not +PLATZ;PLATZ;should +Pseudo ARR-B;CCT;should +Pseudo ARR-B;Response_reg;should +Pseudo ARR-B;tify;should not +RB;RB_B;should +Rcd1-like;Rcd1;should +Rel;RHD_DNA_bind;should +RF-X;RFX_DNA_binding;should +RRN3;RRN3;should +Runt;Runt;should +S1Fa-like;S1FA;should +SAP;STER_AP;should +SBP;SBP;should +SET;zf-C2H2;should not +SET;TCR;should not +SET;CXC;should not +SET;PHD;should not +SET;Myb_DNA-binding;should not +SET;SET;should +Sigma70-like;Sigma70_r2;should +Sigma70-like;Sigma70_r3;should +Sigma70-like;Sigma70_r4;should +Sin3;PAH;should +Sin3;WRKY;should not +Sir2;SIR2;should +SOH1;Med31;should +SRS;DUF702;should +SWI/SNF_BAF60b;SWIB;should +SWI/SNF_SNF2;AP2;should not +SWI/SNF_SNF2;PHD;should not +SWI/SNF_SNF2;SNF2_N;should +SWI/SNF_SNF2;zf-CCCH;should not +SWI/SNF_SNF2;Myb_DNA-binding;should not +SWI/SNF_SNF2;HMG_box;should not +SWI/SNF_SWI3;SWIRM;should +SWI/SNF_SWI3;Myb_DNA-binding;should +TEA;TEA;should +TFb2;Tfb2;should +tify;tify;should +TRAF;BTB;should +TRAF;zf-TAZ;should not +Trihelix;trihelix;should +TUB;Tub;should +ULT;ULT_Domain;should +VARL;VARL;should +VOZ;VOZ_Domain;should +Whirly;Whirly;should +WRKY;WRKY;should +Zinc finger, AN1 and A20 type;zf-AN1;should +Zinc finger, AN1 and A20 type;zf-C2H2;should not +Zinc finger, MIZ type;zf-MIZ;should +Zinc finger, MIZ type;zf-C2H2;should not +Zinc finger, ZPR1;zf-ZPR1;should +Zn_clus;Zn_clus;should +ALOG;ALOG;should +C2H2;C2H2-IDD;should not +C2H2_IDD;C2H2-IDD;should +C2H2_IDD;zf-C2H2;should +MYST;zf-MYST;should +CBP;CBP;should +CBP;zf-TAZ;should +CBP;BTB;should not +TAFII250;DUF3591;should +LOB1;bZIP_1;should not +LOB1;bZIP_2;should not +LOB1;DUF260;should +LOB1;HLH;should not +LOB1;Homeobox;should not +LOB2;LOB2;should +LDL/FLD;SWIRM;should +LDL/FLD;Myb_DNA-binding;should not +ADA2;zz-ADA2;should +ADA2;Myb_DNA-binding;should +RKD;RWP-RK;should +RKD;NLP;should not +NLP;RWP-RK;should +NLP;NLP;should +CRF;CRF;should +CRF;AP2;should +GIY_YIG;GIY_YIG;should +ZPR;ZPR;should +HD-LD;LD;should +HD-NDX;NDX;should +HD-SAWADEE;SAWADEE;should +C1HDZ;C1HDZ;should +C1HDZ;Homeobox;should +C1HDZ;START;should not +C1HDZ;MEKHLA;should not +C2HDZ;C2HDZ;should +C2HDZ;Homeobox;should +C2HDZ;START;should not +C2HDZ;MEKHLA;should not |
| b |
| diff -r 196795831b6a -r c4f865bd101a tapscan_script_v74.pl --- a/tapscan_script_v74.pl Wed Feb 14 13:54:16 2024 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 |
| [ |
| b'@@ -1,916 +0,0 @@\n-#!/usr/bin/perl\n-use strict;\n-use warnings;\n-\n-# Written by Gerrit Timmerhaus (gerrit.timmerhaus@biologie.uni-freiburg.de).\n-# Changes included by Kristian Ullrich, Per Wilhelmsson and Romy Petroll.\n-\n-# Script to extract all detected domains out of a hmmsearch results file and classify the families of all used proteins based on these domains.\n-# The classification depends on a table which contains all known classification rules for the protein families of interest and on specific coverage values defined for every domain.\n-# The script provides three outputs, namely output.1, output.2 and output.3. The output files are tables in ";"-delimited format.\n-# The structure of output.1 is: "sequence ID ; TAP family ; number of classifications ; domains". \n-# Output.3 shares in principle the same structure as output.1, except that subfamilies are considered. ("sequence ID ; TAP family ; Subfamily ; number of classifications ; domains")\n-# The superior TAP family is specified first, followed by the subfamily. If a TAP family has no subfamily, the TAP family is specified first and then a "-". \n-# The structure of output.2 is: "TAP family";"number of detected proteins".\n-# More than one entry for a protein is possible because the classification rules may allow more than one classification.\n-#\n-# The script must be startet with the arguments <hmmsearch output file> <classification rules> <output classifications file> <output family statistics file> <output subfamily classifications file> <"filter" if desired>\n-\n-if (!@ARGV or ($ARGV [0] eq "-h") or ($ARGV [0] eq "-help")) {\n-\tprint "Usage: extract.and.classify.pl <hmmsearch output file> <classification rules> <output classifications file> <output family statistics file> <output subfamily classifications file> <\\"filter\\" (if desired)>\\n\\n";\n-\texit;\n-}\n-\n-# hmmsearch_output: domtblout file\n-my $hmmsearch_output = $ARGV [0];\n-# decision_table: rules file\n-my $decision_table = $ARGV [1];\n-# family_classifications: output.1\n-my $family_classifications = $ARGV [2]; \n-# family_statistics: output.2\n-my $family_statistics = $ARGV [3];\n-# subfamily_classifications: output.3\n-my $subfamily_classifications = $ARGV [4];\n-# domspec_cuts: coverage values file\n-my $domspec_cuts = $ARGV [5];\n-# gene_model_filte: filter for ARATH and ORYSA\n-my $gene_model_filter = $ARGV [6];\n-\n-if ($family_statistics eq "") {\n-\tprint "Usage: extract.and.classify.pl <hmmsearch output file> <classification rules> <output classifications file> <output family statistics file> <output subfamil classifications file> <\\"filter\\" (if desired)>\\n\\n";\n-\texit;\n-}\n-\n-if ($gene_model_filter and $gene_model_filter eq "filter") {\n-\tprint "\\nGene model filter is activated. It only works for TAIR (Arabidopsis) and TIGR (Rice) proteins up to now\\n";\n-}\n-\n-# Array where the $hmmsearch-output/domtblout file will be stored\n-my @output = ();\n-# Array with domain-specific coverage values\n-my @cuts = ();\n-# Array with rules\n-my @dec_table = ();\n-# Counter for the number of detected domains in the hmmsearch output file\n-my $entry_counter = 0;\n-# Containes the actual result for a query sequence\n-my $akt_entry = "";\n-# Used to define query entry to ignore similar domains\n-my $whole_entry = ""; \n-# Includes the final entries after ignoring similar domains\n-my @results_of_extraction = ();\n-# Used to define query entry to ignore similar domains\n-my $extracted_domain = "";\n-# Used to define query entry to ignore similar domains\n-my $present = "";\n-# Used to define query entry to ignore similar domains\n-my $protein = ""; \n-\n-my $lek = "";\n-\n-############################################\n-### 1. Read in the hmmsearch output file ###\n-############################################\n-\n-print "\\n*** reading in $hmmsearch_output ***\\n\\n";\n-\n-@output = get_file_data("$hmmsearch_output");\n-\n-print "*** Parsing $hmmsearch_output ***\\n\\n";\n-\n-# If wrong format exit the program, ninth row from the end\n-if ($output [-9] !~ /^# Program: hmms'..b'ilies into $family_list to create output.2\n-foreach my $fcf_line (@family_classifications_file) {\n-\t$fcf_line =~ /^[^;]+;([^;]+)/;\n-\t#push @family_list, "$1";\n-\tprint FAMILY_CLASSIFICATIONS "$fcf_line";\n-}\n-close (FAMILY_CLASSIFICATIONS);\n-\n-foreach my $fcf_line (@subfamily_classifications_file) {\n-\t$fcf_line =~ /^[^;]+;([^;]+);([^;]+)/;\n-\tif ($2 eq "-") {\n-\t\tpush @family_list, "$1";\n-\t}\n-\telse {\n-\t\tpush @family_list, "$2";\n-\t}\n-\tprint SUBFAMILY_CLASSIFICATIONS "$fcf_line";\n-} \n-\n-close (SUBFAMILY_CLASSIFICATIONS);\n-\n-print "*** calculating the family statistics and write it in $family_statistics ***\\n\\n";\n-\n-##################################\n-### 5. Create the output files ###\n-##################################\n-\n-my $statistics_outputfile = "$family_statistics";\n-\n-unless (open(FAMILY_STATISTICS, ">$statistics_outputfile")) {\n-\tprint "Cannot open file \\"$statistics_outputfile\\" to write to!!\\n\\n";\n-\texit;\n-}\n-\n-# Count the family entries\n-my @output_family_statistics = ();\n-my @gefundene_familien = ();\n-my $family_counter = 1;\n-\n-shift @family_list;\n-@family_list = sort @family_list;\n-\n-\n-my $old_family = "";\n-push @family_list, \'BAD FIX\'; # Makes to loop go through every fam and stops at non fam(BAD FIX).\n-\n-foreach my $line (@family_list) {\n-\tif ($line eq $old_family) {\n-\t\t$family_counter++;\n-\t}\n-\telsif ($old_family ne "") {\n-\t\tpush (@output_family_statistics,"$old_family;$family_counter\\n");\n-\t\t# Add all found families to a list of found families\n-\t\tpush (@gefundene_familien,"$old_family");\n-\t $family_counter=1;\n-\t}\n-\t$old_family = $line;\n-}\n-\n-my %hash = ();\n-\n-# Put all families from the classifictaion ruled and all found families in a hash\n-foreach my $element (@gefundene_familien,@liste_alle_familien) {$hash{$element}++;}\n-\n-# Remove merged families\n-delete $hash{\'GARP_ARR-B_Myb\'};\n-delete $hash{\'GARP_ARR-B_G2\'};\n-delete $hash{\'bZIP1\'};\n-delete $hash{\'bZIP2\'};\n-delete $hash{\'bZIPAUREO\'};\n-delete $hash{\'bZIPCDD\'};\n-delete $hash{\'HRT\'};\n-delete $hash{\'GIY_YIG\'};\n-\n-# Add all not found families from the classification rules to @output_family_statistics \n-# With zero as number of families found \n-\n-foreach my $element (keys %hash) {\n-\tif (($hash{$element} == 1) and ( $element eq "0_no_family_found")) {\n-\tpush (@output_family_statistics,"$element;$unclassified_families\\n");\n-\t}\n-\tif (($hash{$element} == 1) and ($element ne "0_no_family_found")) {\n-\tpush (@output_family_statistics,"$element;0\\n");\n-\t}\n-}\n-\n-# Sort @output_family_statistics caseinsensitive-alphabetically \n-my @sortierte_statistik = sort {lc $a cmp lc $b} @output_family_statistics;\n-\n-# Print headline to @sortierte_statistik\n-unshift (@sortierte_statistik,"family statistics for $hmmsearch_output\\n");\n-\n-print FAMILY_STATISTICS @sortierte_statistik;\n-\n-# Print FAMILY_STATISTICS "$old_family;$family_counter\\n";\n-\n-close (FAMILY_STATISTICS);\n-\n-#################################################\n-### 6. Give out some statistical informations ###\n-#################################################\n-\n-$entry_counter [0] = 0;\n-my $sum = 0;\n-foreach my $entry (@entry_counter) {\n-\t$sum += $entry;\n-}\n-print "$classified_families classifications were found for $sum proteins.\\n";\n-print "This classifications are divided in:\\n";\n-my $count = 0;\n-foreach my $element (@entry_counter) {\n-\tif ($count != 0) {\n-\t\tprint "$element proteins were classified for $count";\n-\t\tif ($count == 1) {print " family\\n";}\n-\t\telse {print " different families\\n";}\n-\t}\n-\t$count++;\n-}\n-print "\\n$unclassified_families proteins could not be classified\\n\\n";\n-\n-print "*** The results were written in $family_classifications and $subfamily_classifications ***\\n";\n-print "*** done ***\\n\\n";\n-\n-exit;\n-\n-sub get_file_data {\n-\t\n-\tmy ($filename) = @_;\n-\n-\tuse strict;\n-\tuse warnings;\n-\n-\tmy @filedata = ();\n-\n-\tunless( open(GET_FILE_DATA, $filename)) {\n-\t\tprint STDERR "Cannot open file \\"$filename\\"n\\n";\n-\t\texit;\n-\t}\n-\n-\t@filedata = <GET_FILE_DATA>;\n-\n-\tclose GET_FILE_DATA;\n-\n-\treturn @filedata;\n-}\n-\n' |
| b |
| diff -r 196795831b6a -r c4f865bd101a test-data/output.2.tsv --- a/test-data/output.2.tsv Wed Feb 14 13:54:16 2024 +0000 +++ b/test-data/output.2.tsv Thu Feb 22 10:07:53 2024 +0000 |
| b |
| @@ -57,13 +57,14 @@ HD-NDX 0 HD-other 0 HD-SAWADEE 0 -HD_BEL 0 HD_DDT 0 -HD_KNOX1 0 -HD_KNOX2 0 HD_PHD 0 HD_PINTOX 0 HD_PLINC 0 +HD_TALE 0 +HD_TALE_BEL 0 +HD_TALE_KNOX1 0 +HD_TALE_KNOX2 0 HD_WOX 0 HMG 0 HSF 0 |
| b |
| diff -r 196795831b6a -r c4f865bd101a test-data/output.domtbl.tsv --- a/test-data/output.domtbl.tsv Wed Feb 14 13:54:16 2024 +0000 +++ b/test-data/output.domtbl.tsv Thu Feb 22 10:07:53 2024 +0000 |
| [ |
| @@ -20,9 +20,9 @@ # Program: hmmsearch # Version: 3.3.2 (Nov 2020) # Pipeline mode: SEARCH -# Query file: /home/saskia/code/github/galaxyproject/tools-iuc/tools/tapscan/tapscan_domains_v12.txt -# Target file: /tmp/saskia/tmp4n93kzh1/files/b/7/b/dataset_b7b39fd6-41f1-440d-bfb7-da7a3a9f070e.dat -# Option settings: hmmsearch --domtblout domtblout.txt --cut_ga /home/saskia/code/github/galaxyproject/tools-iuc/tools/tapscan/tapscan_domains_v12.txt /tmp/saskia/tmp4n93kzh1/files/b/7/b/dataset_b7b39fd6-41f1-440d-bfb7-da7a3a9f070e.dat -# Current dir: /tmp/saskia/tmp4n93kzh1/job_working_directory/000/4/working -# Date: Mon Nov 13 14:57:28 2023 +# Query file: /home/saskia/code/github/bgruening/galaxytools/tools/tapscan/tapscan_domains_v13.txt.gz +# Target file: /tmp/saskia/tmpwta2cyq_/files/f/5/c/dataset_f5c238a3-8cc8-42f9-86b5-ebc2431be570.dat +# Option settings: hmmsearch --domtblout domtblout.txt --cut_ga /home/saskia/code/github/bgruening/galaxytools/tools/tapscan/tapscan_domains_v13.txt.gz /tmp/saskia/tmpwta2cyq_/files/f/5/c/dataset_f5c238a3-8cc8-42f9-86b5-ebc2431be570.dat +# Current dir: /tmp/saskia/tmpwta2cyq_/job_working_directory/000/4/working +# Date: Thu Feb 22 10:10:40 2024 # [ok] |