Mercurial > repos > artbio > sr_bowtie_dataset_annotation

--- a/barplot.r	Mon Apr 11 00:27:41 2022 +0000
+++ b/barplot.r	Tue Nov 15 00:45:36 2022 +0000
@@ -1,13 +1,13 @@
 if (length(commandArgs(TRUE)) == 0) {
-  system("Rscript barplot.r -h", intern = F)
+  system("Rscript barplot.r -h", intern = FALSE)
   q("no")
 }

 # load packages that are provided in the conda env
-options(show.error.messages = F,
+options(show.error.messages = FALSE,
        error = function() {
            cat(geterrmessage(), file = stderr())
-                           q("no", 1, F)
+                           q("no", 1, FALSE)
        }
 )
 loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")
@@ -35,7 +35,7 @@
 opt <- parse_args(OptionParser(option_list = option_list),
                  args = commandArgs(trailingOnly = TRUE))

-annotations <- read.delim(opt$input, header = F)
+annotations <- read.delim(opt$input, header = FALSE)
 colnames(annotations) <- c("sample", "class", "percent_of_reads", "total")
 annotations$percent <- round(annotations$percent_of_reads / annotations$total * 100,
                             digits = 2)
@@ -51,7 +51,7 @@
        geom_bar(position = "fill", stat = "identity") +
        facet_wrap(~sample, ncol = 3) +
        geom_label_repel(aes(label = percent), position = position_fill(vjust = 0.5),
-                        size = 2, show.legend = F) +
+                        size = 2, show.legend = FALSE) +
        coord_polar(theta = "y") +
        labs(x = "Class fractions (%)") +
        scale_fill_manual(values = mycolors) +
--- a/sr_bowtie_dataset_annotation.xml	Mon Apr 11 00:27:41 2022 +0000
+++ b/sr_bowtie_dataset_annotation.xml	Tue Nov 15 00:45:36 2022 +0000
@@ -1,4 +1,4 @@
-<tool id="sr_bowtie_dataset_annotation" name="Annotate smRNA dataset" version="2.7">
+<tool id="sr_bowtie_dataset_annotation" name="Annotate smRNA dataset" version="2.8">
   <description>by iterative alignments with sRbowtie</description>
   <requirements>
         <requirement type="package" version="1.3.1">bowtie</requirement>
@@ -8,14 +8,14 @@
   </requirements>
   <command  detect_errors="exit_code"><![CDATA[
         #if $refGenomeSource1.genomeSource == "history":
-            bowtie-build -f $refGenomeSource1.ownFile genome  1>/dev/null &&
+            bowtie-build --threads \${GALAXY_SLOTS:-4} -f $refGenomeSource1.ownFile genome  1>/dev/null &&
             #set index_path = 'genome'
         #else:
             #set index_path = $refGenomeSource1.index.fields.path
         #end if

         #for $i in $AdditionalQueries:
-            bowtie-build -f $i.ownFile $i.ownFile.name  1>/dev/null &&
+            bowtie-build --threads \${GALAXY_SLOTS:-4} -f $i.ownFile $i.ownFile.name  1>/dev/null &&
         #end for

         #set method_prefix = "-v %s -k 1 --best" % str($mismatches)
@@ -121,11 +121,11 @@
             <output name="output" ftype="tabular" file="multisample5_output.tab" />
             <output name="barplot" ftype="pdf" file="multisample5_output.pdf" compare="sim_size" delta="500" />
             <output_collection name="unmatched" type="list" count="5">
-                <element name="sample5.fa_unmatched" file="unmatched_5.fa" ftype="fasta"/>
-                <element name="sample4.fa_unmatched" file="unmatched_4.fa" ftype="fasta"/>
-                <element name="sample3.fa_unmatched" file="unmatched_3.fa" ftype="fasta"/>
-                <element name="sample2.fa_unmatched" file="unmatched_2.fa" ftype="fasta"/>
-                <element name="sample1.fa_unmatched" file="unmatched_1.fa" ftype="fasta"/>
+                <element name="sample5.fa_unmatched" file="unmatched_5.fa" ftype="fasta" compare="sim_size" delta="0"/>
+                <element name="sample4.fa_unmatched" file="unmatched_4.fa" ftype="fasta"  compare="sim_size" delta="0"/>
+                <element name="sample3.fa_unmatched" file="unmatched_3.fa" ftype="fasta"  compare="sim_size" delta="0"/>
+                <element name="sample2.fa_unmatched" file="unmatched_2.fa" ftype="fasta"  compare="sim_size" delta="0"/>
+                <element name="sample1.fa_unmatched" file="unmatched_1.fa" ftype="fasta"  compare="sim_size" delta="0"/>
             </output_collection>
         </test>
         <test>
@@ -137,7 +137,7 @@
             <output name="output" ftype="tabular" file="sample1_output.tab" />
             <output name="barplot" ftype="pdf" file="sample1_output.pdf" compare="sim_size" delta="500"/>
             <output_collection name="unmatched" type="list">
-                <element name="sample1.fa_unmatched" file="unmatched_1.fa" ftype="fasta"/>
+                <element name="sample1.fa_unmatched" file="unmatched_6.fa" ftype="fasta" compare="sim_size" delta="0"/>
             </output_collection>
         </test>
         <test>
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/unmatched_6.fa	Tue Nov 15 00:45:36 2022 +0000
@@ -0,0 +1,530 @@
+>25207
+TCGCACTAGCTTCGGTCGTCTCGGA
+>25229
+TCTGTTCGAGGTTCTTTATTTGAA
+>25283
+TAAGGGAGCGAGATCGTTCGATGCA
+>25363
+TAGTGTTGGGTGTAGGCCAGGGAG
+>25376
+TCCGCCGACAATGTTGCAACATGA
+>25435
+TTTCTTCAAGCTGCGCGTTTTTCGG
+>25593
+TCGAAAACTTCTTCGCACTCTTCGTT
+>25660
+TCTCGCTGCGTGAACGATGAAGGC
+>25968
+TTTGCTAGGACTGCTGCATAAG
+>26220
+GCCCGTGTCGATGAATGCTTCAAACT
+>26261
+TATTATCGTCCGCTGCTAAACTGC
+>26543
+CTTATGTATAAAAAGCTCTGATGGA
+>26610
+TCCTTAAGTTCGTTGATTTGGGCT
+>26675
+CAATATCGTCAACATCCTCGAACGAT
+>26873
+TTGGGAGCTGAATCCCGTTACGGTA
+>26978
+CGCGCATTAGCTCAAATCTAGTTGGA
+>27220
+TGTTTATTTGTCAAGTTTAGATAATA
+>27397
+TAGTTCACAGCTGTATGTCCAGATGGGT
+>27869
+TATGGTCCAGAATGTAGCCTCGGC
+>27878
+TATTAATCGGGCCACAAATATCGGTA
+>28035
+GGTATTCTTTGCGAGGTCGTCCTGG
+>28154
+AAAGACGAGAACGCGTATATGTGTGC
+>28319
+TGACCAACCTTAAAAGATCGGGGT
+>28387
+TATTACTATTTCTAAGCTTTGTTTAAA
+>28594
+TTTTGAGGTTGGTCAAGAAGTTGTT
+>28608
+TAAGATTGAAAATTACTGTGGAGT
+>28668
+TACAGATTCTAGAGACAAAGACGC
+>28674
+TAAGTACATGCGCCCAGCCGCCGTGA
+>28836
+TCTGGTTAAGGTCGGAATACTCGTCT
+>29056
+TGCTTTACATACCCTTTGGTGCCC
+>29323
+TTTTGCTGCAGAGCTTCTTCCAACGT
+>29342
+TATGTATGGATATGTATATTTATGGT
+>29387
+TGATAATCGACCTCTTCCATCGTTGT
+>29423
+TCTAAGAACTTCTGAGGTGAAGG
+>29462
+TGCATTTCAATCGGAAGAGTACTCTG
+>29492
+TAAAACACAAATCTCGACATACAGA
+>29703
+GTACAGGTTCTGATGACAATG
+>29785
+TCAGATGAAAGACAAATTAGAATT
+>30053
+TAAGAACTTCTGAGATGAAGGGC
+>30096
+TGGATATTGAATGTTTTTGATTTGC
+>30136
+TGTAGTCGTCGTATGTCCGGA
+>30272
+TGCATTCGTGGATTCGCATTCGAGA
+>30497
+TGACAAAATATGCCCTTCACCTCAGA
+>30787
+TAACTGTTTTTAGTTCAAAGTCTCGGA
+>30836
+TAAATATTTTTTTTGAAACA
+>30977
+CGACTGTAATTATTAGCACAATACT
+>30985
+GACGATATTGCTGCAATAGACCTTGA
+>31000
+TCAGATGAGAGACAAATTAGAAT
+>31256
+TAATTCGGAATGCCTGCTCTACT
+>31417
+TCAATGATCGCTGTGCTCAGTAGGA
+>31506
+TTTGTCTGACGTTAAAAAATATA
+>31567
+TCCACATTAGGAGGATTATTAGACAAC
+>31790
+AAACATAATAATTGATGGCGGAAGA
+>31872
+AAGGTAATCATAGAGCACCACGGTT
+>32157
+TTTCTGTGAATTCACATGCTGATGA
+>32192
+TTTCTCATGTATAAAATGCTCTGATGG
+>32223
+TATCTTGTTATTCTAGTGTCTTTGGTT
+>32338
+TGTGGGACTCGAGCCAAAATGGCAACCT
+>32497
+TGCGGTTGGACAATTTTTTTTTTATA
+>32506
+TTGTTGTTTGGAGGAAGTTCCTTT
+>32510
+TCTTCCGCCATCAATTATTATGTTTT
+>32522
+TACTTGACTTTTCTATAGAATCTGGT
+>32540
+TATCGTCAACATCCTCGAACGATCGAGA
+>32626
+TTGATCAAGGTAGGGTTGTCGC
+>32646
+TCTGTTAAACACCCCTGAATCGTGGAT
+>32657
+TTTGGACATTTTGCAGGTGATACAAT
+>32682
+TAACTGTTTTTAGTTCAAAGTCTCGGA
+>32716
+TATCTATAGTTCCGATTGGCCATCTC
+>32885
+GAAAGTGGGTATCTGTATTTTAGGC
+>32967
+TAAAGATACCATCTAACCTCCTTGGA
+>33075
+ACAGATATTAGTGTTTTTCAAGCAGC
+>33147
+ATCGCAATTATGGCATAACAGATTCGGA
+>33163
+TTCAGCGTCGCTTGATTGAATAGAT
+>33231
+TAAGAACTTCTGAGGTGAAGGGC
+>33242
+TTGGGTTTAGAAATTAAAATTAAGGC
+>33244
+TATAGAAAGTTACGAATATATTAGGA
+>33248
+TTTTTGATCAATTGGCACCGTGCGAA
+>33296
+TGCAGAGCTTCTTCCAACGTTGGCAAG
+>33318
+TAGATGTCTGCAGGAATAACGGA
+>33338
+TCGACTATTAATGGCTGTTAGAATT
+>33475
+CAAACTTATCGACCATCTCCTCAAACG
+>33601
+TTACCCTTTTTCCGGAGCGTTTGTGC
+>33611
+AATGATCGCTGTGCTCAGTATGACGG
+>33662
+TTTGATTCATTACAATTTACGCTGAA
+>33737
+CAGATATTAGTGTTTTTCAAGCAGC
+>33747
+TACATGTAAAGCAGCTGTGTGTGC
+>33815
+AGTTTTTGGAATCACTTGA
+>34032
+TGCGGACGTGTGCTCGCTGCGTGA
+>34200
+TAATATATGTATAATCTGCTTGGTG
+>34240
+TCTTTACGCCATATAAATCATTTCGA
+>34259
+ACAAATCATAAATTTGATGGGACGA
+>34268
+TCTTGTGGACTTCACTCTAG
+>34297
+TAAGTAAATAGTCCCCGCCTTATTGAGG
+>34337
+TGGATAATGAATGTTTTTGATTTGC
+>34569
+TCTAGGTTCTTCTGAAATCGTGGGA
+>34643
+TCGTGTAGACCGGATAAGATTTTTT
+>34697
+TCTAATACTGTGAAAGGGTGGGG
+>34808
+GTAGGATGTGCTCTGCGGTTTCCAC
+>34892
+TCTGTCGCAGTTGTAGCTTGCAATA
+>34896
+TGTAGTTGCCACTTATGCTGTCCA
+>35151
+TAGCAATGTCCGTCTGTCCGTATGA
+>35196
+TGTAAATGGTCAGCGAAAGCAAAGG
+>35211
+CAGATATTAGTGTTTTTCAAGCAGCGG
+>35246
+CCTCAGAGAACGTCAGACCGCG
+>35272
+TGCAGAGCTTCTTCCAACGTTGG
+>35273
+TCTGGTAGTAAGAAAAATGTAGCTT
+>35286
+TATGTATGGATATGTATATTTATGGGT
+>35317
+TTTTGGTTTGATCGTCAGGTGGTC
+>35512
+GACGATAATAGTGAATTTTGGACA
+>35566
+TTCGAATTCGCGCCATTTCACAATC
+>35948
+TGGCCTGTATACGCTTTCTGTTG
+>36009
+TTGCAAAAGTCATATCTTGAGG
+>36059
+TAAGTTACTATGGATCCATAAGGGTA
+>36100
+TGATATGGGACTTGTAGCTTTTTTAAA
+>36121
+TGTCGTGGGCTGTGCGTTTGAGAA
+>36235
+TGCTACCTTTAGCTGCAAGATTAACT
+>36238
+CAGATATTAGTGTTTTTCAAGCAGCGG
+>36523
+TGCGAGTCCGAGCTTTGAACGTGGG
+>36607
+TCTCAATGTAATGTCTTCTTTTTGGA
+>36624
+CGTTCTTTAAAACCACCAATGGGA
+>36776
+TACTCACGACCATTGTCATTTCTCAAG
+>36825
+CGTTCGAGGTCCACTTTCTTAGCGGA
+>36850
+TCTGATGTCGGCGAGAAAGGAGTCTCA
+>36872
+TCAAGGCCAGCACACAGTAACATGGT
+>36968
+ACATTGGAAATACCGCGGGACCGC
+>36994
+TATACGTTTTTTGACCTCTTCTCTTTGA
+>37110
+TGATAATGTAAACAAAGATAAAGGG
+>37210
+TCAGATGAAAGACAAATTAGAATTAAA
+>37224
+GACGATAATAGTGAATTTTGGACA
+>37226
+TCTGAAGTTGCCGCACTAGAGATGG
+>37339
+TATACAGTCCACTATATCGTTGTTTAA
+>37395
+TACAATTTACGCTGAATTTAAATGAA
+>37460
+CACAAAGTAACGTGCACCACCATTT
+>37766
+TCTGTGCAACTTTGTACGCGAAGTCAGC
+>37836
+TCAAGGATTAATGTAGGGGGGGGG
+>38096
+TTTGGAGGTAGAAGTCTTAGTGGCCGC
+>38100
+GACAAAATATGCCCTTCAATTTAGA
+>38113
+TATGGTCCAGAATGTAGCCTCGGC
+>38213
+TAACTGTTTTTAGTTCAAAGTCTCGAA
+>38224
+TTTCAGTTTTTATTGTTAGTCACAGG
+>38263
+TAACTGCGAAGTCGATCAGGTCCGA
+>38266
+ATCCGGACGATTGACGAGGAGCCCATT
+>38271
+TTATGTCAGTGTCGAAGGCGATCGAA
+>38373
+TCTATAGCCTTGGCGTAGGAACTCGCA
+>38384
+GCTACAACTGCGACAGAAAATTCGGA
+>38468
+TATTATCGTCCGCTGCTAAACTGC
+>38491
+TGCATACTTCCGTTCTCTTTTCGGGA
+>38673
+TTGTGTATTGTGATTCTGATTCGTG
+>38745
+TCCCTGGCCCGCTAGACAGCAGGA
+>38784
+TGTGCTAATAATTAAAGTCGACTGA
+>38873
+TGTCATACTTTCGTCAAAA
+>39154
+TAATAGTCAGGGCGCGAATTTTTAAAA
+>39325
+TTAGGACTTATTGAACTTTACGGTA
+>39359
+TATGCAAATCAAGTGTGACCGTAGCT
+>39450
+TTAAACAACGATATAGTGGACAGTA
+>39453
+TCTGCCGGATTTTGATCCAATCAAGG
+>39819
+GTGGAATTGAAAAAGAACCAGACACA
+>40040
+TTTTTGGAACTACCTGAGTCGGTT
+>40183
+TAACACAAAGCAGTATGATTTAATAAT
+>40189
+CAGCAAGCTGAGATGTACATTAGTATA
+>40374
+TACGTTTTCTTGCAGATCAAAAA
+>40445
+CGTTCTTTAAAACCACCAATGGGA
+>40467
+TACGCAGATTCCTGGGAGTTACAGGA
+>40488
+TGATTTGGGCTTGCATACTTGTACT
+>40797
+TATATTCGTGTTCATGTGTGAACAGC
+>40831
+TCTAAGAACTTCTGAGGTGAAGGGC
+>40951
+TTCTTCGTAAGTCAAAATAGTGTCGCC
+>40988
+TATGATTGATTGCTTGAGAGT
+>40998
+TGATAGAGCTGCATTTGAATTAACGG
+>41061
+CTCTTTCCGCTCACTCCCGCTGAGA
+>41076
+TGCATTAAGAAGATTTAGGATCC
+>41140
+TCAAGGATTAATGTAGGGGGGG
+>41143
+CAATAGCGTCGCTGAGTAACAGTG
+>41162
+TACGGATTGCAGCGGCTAG
+>41183
+TTTTTTGGCACACGATTTTTTGGACGT
+>41227
+TAATATATGTATACTCTGCTTGGTG
+>41265
+TTTATGATTTTTGGTAATA
+>41365
+TAGGAGGGTTCCACAACTATTTCGGGG
+>41837
+TAGTTTGACACTGTTTGGAGACGTGG
+>42053
+TCTGTGGTCGAATCGAAGGAGTGC
+>42113
+TAATAGATCGCTCACCTGTTCCTGG
+>42392
+CGGCATCGGAAAACTCCCAGCGGGGC
+>42504
+TAACGTTATTATTATTTGAAAATAGAA
+>42566
+TTTGTTGGGTAGGAACTTTACTGC
+>42668
+TGCTGCAGAGCTTCTTCCAACGTTGG
+>42751
+CCAAAGTCTGGTTGTCAGAAAATGTGC
+>42777
+AAGGAAGGAACCAAAGAAGCACAAACG
+>42804
+TAATTCTAATTTGTCTCTCATCTGA
+>42830
+TAGTATACTTATTAAGTCATTTGA
+>42857
+AGAGTATTCATCTTGAGGCGTGTG
+>42886
+TGAACAACGATTTATGTATATAAGAA
+>42899
+TACGATAATAGTGAATTTTGGACA
+>42936
+TTTGACTAAAAGTCGCTTGTTTTGGA
+>43050
+AATACATAACTCTGGACACAGGAGA
+>43219
+TACTTTCGTCAAAATGTTCAGGAGCT
+>43242
+TGACATGTCTATTTCCATGGGTTCGGA
+>43268
+GTGTATAAAAAAATTTATTGTTGAGCA
+>43299
+TCTGGATGATGGCTGATGCTCGTTG
+>43491
+TCTGATGACAATGAATTTTTTAGACA
+>43520
+TATTGTTGTAATTGCTGCCTCGGTTG
+>43654
+ATATGAACAAAGCAAAGACACTAGAA
+>43677
+CACGATTCATAACCCTCAGCTGAAG
+>43776
+GCTCCTGGCAACTCTGTGATGGAC
+>43903
+TATTTTGATGTTTAATGAA
+>43990
+TTTCTACTTCGTATTATTTTTATGA
+>44062
+AAACATAATAATTGATGGCGGAAGA
+>44071
+CACAGACGCAGTGGAAACCGCAGA
+>44078
+CGTGCGTCCGAGCAAAAGGTGGT
+>44187
+TTATGTAAGAATATTTGTCATTAGA
+>44239
+TATCAATGTTGACCGTAATACTCAA
+>44253
+TGGATATTGAATGTTTTTGATTTGC
+>44279
+TGTGATTTTCCCAATTTATATTAATACA
+>44374
+TAGTCGGAGTTGATGAGCTGCC
+>44412
+TCCGAATTTTCTGTCGCAGTTGTAGCT
+>44423
+TCGGCTTGGGTTTAGAAATTAAAAT
+>44558
+ACAAATCATAAATTTGGTGGGA
+>44637
+TTTTTTATCAATTGGCACCATGCGAA
+>44642
+TACTGTGTGCTGGCCTTGATGAAAGT
+>44688
+TAGATGTCTGCACGAATAACGGA
+>44945
+TGCTTCCGAGCAATCTACGTTGGTAAAA
+>45054
+TCTCATCTGACAATTTTTTAAAAGCGA
+>45088
+TCTGAAGCAGCGCTCACGGCAGAATGC
+>45203
+TGCCGGATTTTGATCCAATCAAGGGA
+>45263
+TATTGATTTTCCTATTTAGTTGAACA
+>45274
+AAAAGTCTGGATATTGTAGGATAGGA
+>45358
+CAACGCTGGACCTTGGACTCGAGGGC
+>45396
+CATCTTGTTATTCTAGTGTCTTTGGTA
+>45641
+GAACTAAGATCAGTCGACTGTAATTAT
+>45774
+CGACGAACTAGCAGCTCTGGTGT
+>45911
+CAGATATTAGTGTTTTTCAAGCAGC
+>45997
+TTAACCAGTCGGCGTTGTTTAAGTAGC
+>46031
+AAAGCGTCTACTTGAACAATGAGA
+>46054
+TTAGATCGTATTACTTGGGTGCTGG
+>46199
+TAAGCGTTAGGTCGTATTACTTGGGC
+>46255
+TAAGATTGAAAATTACTGTGGAGTAAT
+>46366
+TAAGGAATTGTCGGCCATTTAATGTGA
+>46387
+CAGAGCTTCTTCCAACGTTGGCAAG
+>46701
+TTGTATCTTTTTGCTTTTTATATT
+>46705
+TAACTGTTTTTAGTTCAAAGTCTCGGA
+>46783
+TCTCATCTGACAATTTTTTAAAAGCGA
+>46836
+TCATAAGGACAGACGGACAT
+>46892
+AAACATAATAATTGATGGCGGAAGA
+>47244
+TCTCTGTCCGCTCGCTTACGATGAGA
+>47290
+TTTCAGGAATGGGGTCGTCCCACTA
+>47348
+TGCAAAAGTAATATCAAAGACACTA
+>47359
+TTTAAAATTATAATAGTCAGGG
+>47427
+CATCTTGTTATTCTAGTGTCTTTGGT
+>47654
+TTAAACACTGAATTCGGTTTCGAAA
+>47656
+TTTCTGAAGAATCCTGTAACTCCC
+>47785
+TGTAGATGAGCGGCAAATGTGG
+>47827
+TTGATCAAGGTAGAGTTGTCGCGC
+>48184
+TTCAAGGATTAATGTAGGGGGGG
+>48576
+TTAACCCGGAGACTTGGGTGTGGGT
+>48764
+TGCAAAAGTAATATCAAAGACAATAGA
+>48782
+TGACAATGTAGTGAACGCCAGTGT
+>48893
+TATAAATGCCGTCTGATATTATTAAA
+>49017
+ACCGGATGTCATATCCAGCGTCGTGAA
+>49381
+TACAATGTAAATTCGTTTCTTCGATCA
+>49456
+TTTTGGTTTGATCGTCAGGTGGACGC
+>49484
+TCTGCTATCATTGACTCGATCATTGA
+>49569
+TCGATTGTATGATCAGTGGAAGTGGC
+>49602
+GAAATTGGCCAACATTAATTCGGAA
+>49628
+TACTTTCGTCAAAATGTTCAGGAGC
+>49715
+TTTGTCCGGGTGCTTCGAAAGAACTCT
+>49778
+CAATAGCGTCGCTGAGTAACAGTG