Mercurial > repos > fubar > blasttools_search_test

comparison blast_tools_search/blasttoolssearch.xml @ 8:186734f1d63c draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Replace plotly_blast_tool content with blasttools_search. :(
author fubar
date Fri, 04 Aug 2023 01:57:51 +0000
parents
children
comparison
equal deleted inserted replaced
7:bb99f2c0e358 8:186734f1d63c
1 <tool name="blasttoolssearch" id="blasttoolssearch" version="3.0">
2 <!--Source in git at: https://github.com/fubar2/galaxy_tf_overlay-->
3 <!--Created by toolfactory@galaxy.org at 04/08/2023 10:36:33 using the Galaxy Tool Factory.-->
4 <description>Runs a legacy Java jar called blasttools from https://github.com/schmidda/blast-tools/tree/master</description>
5 <requirements>
6 <requirement version="0.26.0" type="package">csvtk</requirement>
7 <requirement version="11.0.13" type="package">openjdk</requirement>
8 </requirements>
9 <stdio>
10 <exit_code range="1:" level="fatal"/>
11 </stdio>
12 <version_command><![CDATA[echo "3.0"]]></version_command>
13 <command><![CDATA[bash $runme "$blastn_search_outputs" "$__tool_directory__/BlastTools.jar" "$summary_viruses_viroids" "$all_blasttools_output"]]></command>
14 <configfiles>
15 <configfile name="runme"><![CDATA[#raw
16
17 ## eResearch Office, QUT
18 ## Created: 31 March 2021
19 ## Last modified: 28 September 2022
20 ## Script: Processes Galaxy Australia generated blastN outputs to summarise and report hits to REGULATED and ENDEMIC viruses/viroids.
21 ## Usage: ./run_VirReport_Summary.sh
22 ## changed to accept a single input file name passed as $1
23 ## Ross Lazarus for a ToolFactory wrapper for Robert Barrero
24 ## July 18 2023
25
26
27 # Requirement: One or more GA-VSD .tabular outputs need to be in the folder where the command above (Usage)is executed.
28 # The script will Look for all files with the suffix *.tabular
29
30 #Processing tabular files
31 file=$1
32 var=$(basename $file)
33
34 #STEP1: modify the columns of Galaxy Australia (GA) blast output to the expected format by the BlastTools.jar tool
35 ###### namely: qseqid sgi sacc length pident mismatch gapopen qstart qend qlen sstart send slen sstrand evalue bitscore qcovhsp stitle staxids qseq sseq sseqid qcovs qframe sframe
36 cat $file |csvtk cut -H -t -f 1,19,20,4,3,5,6,7,8,17,9,10,18,22,11,12,24,21,25,15,16,2,23,13,14 | sed 's/ /_/g' > ${var}_all.txt
37 java -jar $2 -t blastn ${var}_all.txt
38 cat summary_${var}_all.txt | grep "virus\|viroid\|endo" > $4
39
40
41 #STEP0: fetch Top 1 Hits
42 cat $file | awk '{print $1}' | sort | uniq > ${var}.top1.ids
43 for i in `cat ${var}.top1.ids`
44 do
45 echo "fetching top hits..." $i 1>&2 ;
46 grep $i $file | head -1 >> ${var}.top1Hits.txt;
47 done
48
49 #STEP1: modify the columns of Galaxy Australia (GA) blast output to the expected format by the BlastTools.jar tool
50 ###### namely: qseqid sgi sacc length pident mismatch gapopen qstart qend qlen sstart send slen sstrand evalue bitscore qcovhsp stitle staxids qseq sseq sseqid qcovs qframe sframe
51 cat ${var}.top1Hits.txt |csvtk cut -H -t -f 1,19,20,4,3,5,6,7,8,17,9,10,18,22,11,12,24,21,25,15,16,2,23,13,14 | sed 's/ /_/g' > ${var}.txt
52
53 #STEP2: summarise the GA blastN files
54 java -jar $2 -t blastn ${var}.txt
55 #filter virus/viroid/endo
56 cat summary_${var}.txt | grep "virus\|viroid\|endo" > summary_${var}_filtered.txt
57
58 #STEP3: fetch unique names from Blast summary reports
59 cat summary_${var}_filtered.txt | awk '{print $7}' | awk -F "|" '{print $2}'| sort | uniq | sed 's/Species://' > ${var}_uniq.ids
60
61 #STEP4: retrieve the best hit for each virus/viroid
62 echo "processing top hits ..." 1>&2
63 touch ${var}_filtered.txt
64 for id in `cat ${var}_uniq.ids`
65 do
66 #print on the screen the name of the virus/viroids to search
67 #echo "fetching species matches ..." $id 1>&2
68
69 #fetch the virus name on the summary_blastn file by selecting the longest alignment (column 3) and highest genome coverage (column 5)
70 grep $id summary_${var}.txt | sort -k3,3nr -k5,5nr | head -1 >> ${var}_filtered.txt
71 done
72
73 #print the header of the inital summary_blastn file
74 cat summary_${var}.txt | head -1 > header
75 #report 1
76 cat header ${var}_filtered.txt > $3
77
78
79
80
81
82 #end raw]]></configfile>
83 </configfiles>
84 <inputs>
85 <param name="blastn_search_outputs" type="data" optional="false" label="blastn_search_outputs" help="Nucleotide blast search output from a Galaxy blast search" format="tabular" multiple="false"/>
86 </inputs>
87 <outputs>
88 <data name="summary_viruses_viroids" format="tabular" label="summary_viruses_viroids" hidden="false"/>
89 <data name="all_blasttools_output" format="tabular" label="all_blasttools_output" hidden="false"/>
90 </outputs>
91 <tests>
92 <test>
93 <output name="summary_viruses_viroids" value="summary_viruses_viroids_sample" compare="diff" lines_diff="0"/>
94 <output name="all_blasttools_output" value="all_blasttools_output_sample" compare="diff" lines_diff="0"/>
95 <param name="blastn_search_outputs" value="blastn_search_outputs_sample"/>
96 </test>
97 </tests>
98 <help><![CDATA[
99
100 **What it Does**
101
102 Wraps https://github.com/schmidda/blast-tools/tree/master as a Galaxy tool as a demonstration for Roberto Barrero
103
104
105
106 ------
107
108
109 Script::
110
111 ## eResearch Office, QUT
112 ## Created: 31 March 2021
113 ## Last modified: 28 September 2022
114 ## Script: Processes Galaxy Australia generated blastN outputs to summarise and report hits to REGULATED and ENDEMIC viruses/viroids.
115 ## Usage: ./run_VirReport_Summary.sh
116 ## changed to accept a single input file name passed as $1
117 ## Ross Lazarus for a ToolFactory wrapper for Robert Barrero
118 ## July 18 2023
119 # Requirement: One or more GA-VSD .tabular outputs need to be in the folder where the command above (Usage)is executed.
120 # The script will Look for all files with the suffix *.tabular
121 #Processing tabular files
122 file=$1
123 var=$(basename $file)
124 #STEP1: modify the columns of Galaxy Australia (GA) blast output to the expected format by the BlastTools.jar tool
125 ###### namely: qseqid sgi sacc length pident mismatch gapopen qstart qend qlen sstart send slen sstrand evalue bitscore qcovhsp stitle staxids qseq sseq sseqid qcovs qframe sframe
126 cat $file |csvtk cut -H -t -f 1,19,20,4,3,5,6,7,8,17,9,10,18,22,11,12,24,21,25,15,16,2,23,13,14 | sed 's/ /_/g' > $ {var}_all.txt
127 java -jar $2 -t blastn $ {var}_all.txt
128 cat summary_$ {var}_all.txt | grep "virus\|viroid\|endo" > $4
129 #STEP0: fetch Top 1 Hits
130 cat $file | awk '{print $1}' | sort | uniq > $ {var}.top1.ids
131 for i in `cat $ {var}.top1.ids`
132 do
133 echo "fetching top hits..." $i 1>&2 ;
134 grep $i $file | head -1 >> $ {var}.top1Hits.txt;
135 done
136 #STEP1: modify the columns of Galaxy Australia (GA) blast output to the expected format by the BlastTools.jar tool
137 ###### namely: qseqid sgi sacc length pident mismatch gapopen qstart qend qlen sstart send slen sstrand evalue bitscore qcovhsp stitle staxids qseq sseq sseqid qcovs qframe sframe
138 cat $ {var}.top1Hits.txt |csvtk cut -H -t -f 1,19,20,4,3,5,6,7,8,17,9,10,18,22,11,12,24,21,25,15,16,2,23,13,14 | sed 's/ /_/g' > $ {var}.txt
139 #STEP2: summarise the GA blastN files
140 java -jar $2 -t blastn $ {var}.txt
141 #filter virus/viroid/endo
142 cat summary_$ {var}.txt | grep "virus\|viroid\|endo" > summary_$ {var}_filtered.txt
143 #STEP3: fetch unique names from Blast summary reports
144 cat summary_$ {var}_filtered.txt | awk '{print $7}' | awk -F "|" '{print $2}'| sort | uniq | sed 's/Species://' > $ {var}_uniq.ids
145 #STEP4: retrieve the best hit for each virus/viroid
146 echo "processing top hits ..." 1>&2
147 touch $ {var}_filtered.txt
148 for id in `cat $ {var}_uniq.ids`
149 do
150 #print on the screen the name of the virus/viroids to search
151 #echo "fetching species matches ..." $id 1>&2
152 #fetch the virus name on the summary_blastn file by selecting the longest alignment (column 3) and highest genome coverage (column 5)
153 grep $id summary_$ {var}.txt | sort -k3,3nr -k5,5nr | head -1 >> $ {var}_filtered.txt
154 done
155 #print the header of the inital summary_blastn file
156 cat summary_$ {var}.txt | head -1 > header
157 #report 1
158 cat header $ {var}_filtered.txt > $3
159
160 ]]></help>
161 <citations>
162 <citation type="doi">10.1093/bioinformatics/bts573</citation>
163 </citations>
164 </tool>
165