Repository 'rna_starsolo'
hg clone https://toolshed.g2.bx.psu.edu/repos/iuc/rna_starsolo

Changeset 9:ec9cbd6b9a49 (2021-01-15)
Previous changeset 8:00fbfac99d39 (2020-12-04) Next changeset 10:a6fba3d92531 (2021-03-15)
Commit message:
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/rgrnastar commit 00c545ddbf0f008903f4b4c11d476e6089c3f531"
modified:
macros.xml
rg_rnaStarSolo.xml
added:
test-data/indrop.R1.fastq.gz
test-data/indrop.R2.fastq.gz
test-data/indrop.barcodes1.txt
test-data/indrop.barcodes2.txt
test-data/smartseq.cellids.txt
test-data/smartseq1.R1.fastq.gz
test-data/smartseq1.R2.fastq.gz
test-data/smartseq2.R1.fastq.gz
test-data/smartseq2.R2.fastq.gz
test-data/smartseq3.R1.fastq.gz
test-data/smartseq3.R2.fastq.gz
test-data/smartseq4.R1.fastq.gz
test-data/smartseq4.R2.fastq.gz
test-data/smartseq5.R1.fastq.gz
test-data/smartseq5.R2.fastq.gz
test-data/smartseq6.R1.fastq.gz
test-data/smartseq6.R2.fastq.gz
test-data/smartseq7.R1.fastq.gz
test-data/smartseq7.R2.fastq.gz
test-data/smartseq8.R1.fastq.gz
test-data/smartseq8.R2.fastq.gz
test-data/smartseq9.R1.fastq.gz
test-data/smartseq9.R2.fastq.gz
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 macros.xml
--- a/macros.xml Fri Dec 04 22:05:36 2020 +0000
+++ b/macros.xml Fri Jan 15 17:39:11 2021 +0000
[
@@ -5,7 +5,7 @@
     the index versions in sync, but you should manually adjust the +galaxy
     version number. -->
     <!-- STAR version to be used -->
-    <token name="@VERSION@">2.7.6a</token>
+    <token name="@VERSION@">2.7.7a</token>
     <!-- STAR index version compatible with this version of STAR
     This is the STAR version that introduced the index structure expected
     by the current version.
@@ -33,7 +33,7 @@
             <edam_operation>operation_0292</edam_operation>
         </edam_operations>
     </xml>
-    
+
     <xml name="index_selection" token_with_gene_model="0">
         <param argument="--genomeDir" name="genomeDir" type="select"
         label="Select reference genome"
@@ -132,6 +132,35 @@
         #end if
         #end if
         ]]></token>
+    <token name="@READSHANDLING@" ><![CDATA[
+    ## Check that the input pairs are of the same type
+    ## otherwise STARsolo will run for a long time and then error out.
+    ## We consume either repeats of two inputs R1 + R2
+    ## or a collection of paired reads.
+    #if str($sc.input_types.use) == "repeat":
+        #set $reads1 = []
+        #set $reads2 = []
+        #for $r1, $r2 in zip($sc.input_types.input1, $sc.input_types.input2):
+            #assert $r1.datatype == $r2.datatype
+            #silent $reads1.append(str($r1))
+            #silent $reads2.append(str($r2))
+        #end for
+        #set $reads1 = ','.join($reads1)
+        #set $reads2 = ','.join($reads2)
+    #elif str($sc.input_types.use) == "list_paired":
+        #set $r1 = $sc.input_types.input_collection.forward
+        #set $r2 = $sc.input_types.input_collection.reverse
+        #set $reads1 = $r1
+        #set $reads2 = $r2
+    #end if
+    ## cDNA sequence(s) [R2] always go first, then barcode(s) [R1]
+    ## see: Section 3.2 of STAR manual for multiple inputs, and Section 13 for STARsolo inputs
+    --readFilesIn $reads2 $reads1
+    --soloCBmatchWLtype $sc.soloCBmatchWLtype
+    #if $r1.is_of_type('fastq.gz', 'fastqsanger.gz'):
+        @FASTQ_GZ_OPTION@
+    #end if
+    ]]></token>
     <xml name="ref_selection">
         <param argument="--genomeFastaFiles" type="data" format="fasta" label="Select a reference genome" />
         <!-- Currently, this parameter is not exposed in the wrapper,
@@ -148,4 +177,53 @@
             <yield />
         </stdio>
     </xml>
+    <xml name="input_selection">
+        <conditional name="input_types" >
+            <param name="use" type="select" label="Input Type" >
+                <option value="repeat" >Separate barcode and cDNA reads</option>
+                <option value="list_paired" >Paired collection of barcode and cDNA reads</option>
+            </param>
+            <when value="repeat">
+                <param format="fastq,fasta,fastq.gz,fastqsanger.gz" name="input1" type="data"  multiple="true"
+                label="RNA-Seq FASTQ/FASTA file, Barcode reads" />
+                <param format="fastq,fasta,fastq.gz,fastqsanger.gz" name="input2" type="data"  multiple="true"
+                label="RNA-Seq FASTQ/FASTA file, cDNA reads"/>
+            </when>
+            <when value="list_paired">
+                <param name="input_collection" collection_type="paired" type="data_collection" format="fastq,fasta,fastq.gz,fastqsanger.gz" label="Collection of Pairs" />
+            </when>
+        </conditional>
+    </xml>
+    <xml name="input_selection_smart_seq">
+        <conditional name="input_types_smart_seq" >
+            <param name="use" type="select" label="Input Type" >
+                <option value="list_single_end" >Single-end FASTQ collection</option>
+                <option value="list_paired_end" >Paired FASTQ collection</option>
+            </param>
+            <when value="list_single_end">
+                <param name="single_end_collection" collection_type="list" type="data_collection" format="fastq,fasta,fastq.gz,fastqsanger.gz" label="List of single-end FASTQ files" />
+            </when>
+            <when value="list_paired_end">
+                <param name="paired_end_collection" collection_type="list:paired" type="data_collection" format="fastq,fasta,fastq.gz,fastqsanger.gz" label="List of paired-end FASTQ files" />
+            </when>
+        </conditional>
+    </xml>
+    <xml name="umidedup_options">
+        <option value="1MM_All" selected="true">All</option>
+        <option value="1MM_Directional" >Directional</option>
+    </xml>
+    <xml name="anchor_types">
+        <option value="0">Read start</option>
+        <option value="1">Read end</option>
+        <option value="2">Adapter start</option>
+        <option value="3">Adapter end</option>
+    </xml>
+    <xml name="cb_match_wl_common">
+        <option value="Exact" >Exact</option>
+        <option value="1MM" >Single match</option>
+    </xml>
+    <xml name="cb_match_wl_cellranger">
+        <option value="1MM_multi" selected="true" >Multiple matches (CellRanger 2)</option>
+        <option value="1MM_multi_pseudocounts" >Multiple matches (CellRanger 3)</option>
+    </xml>
 </macros>
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 rg_rnaStarSolo.xml
--- a/rg_rnaStarSolo.xml Fri Dec 04 22:05:36 2020 +0000
+++ b/rg_rnaStarSolo.xml Fri Jan 15 17:39:11 2021 +0000
[
b'@@ -17,68 +17,89 @@\n     STAR\n     @REFGENOMEHANDLING@\n \n-    --readFilesIn\n-    ## Check that the input pairs are of the same type\n-    ## otherwise STARsolo will run for a long time and then error out.\n-    ## We consume either repeats of two inputs R1 + R2\n-    ## or a collection of paired reads.\n+    ## Supports Drop-seq, 10X Chromium, inDrop and Smart-Seq\n+    --soloType $sc.solo_type\n \n-    #if str($input_types.use) == "repeat":\n-        #set $reads1 = []\n-        #set $reads2 = []\n-        #for $r1, $r2 in zip($input_types.input1, $input_types.input2):\n-            #assert $r1.datatype == $r2.datatype\n-            #silent $reads1.append(str($r1))\n-            #silent $reads2.append(str($r2))\n-        #end for\n-        #set $reads1 = \',\'.join($reads1)\n-        #set $reads2 = \',\'.join($reads2)\n-    #elif str($input_types.use) == "list_paired":\n-        #set $r1 = $input_types.input_collection.forward\n-        #set $r2 = $input_types.input_collection.reverse\n-        #set $reads1 = $r1\n-        #set $reads2 = $r2\n-    #end if\n-\n-    ## cDNA sequence(s) [R2] always go first, then barcode(s) [R1]\n-    ## see: Section 3.1 of STAR manual for multiple inputs, and Section 13 for STARsolo inputs\n-    $reads2 $reads1\n-\n-    #if $r1.is_of_type(\'fastq.gz\', \'fastqsanger.gz\'):\n-        @FASTQ_GZ_OPTION@\n-    #end if\n-\n-    ## Droplet is the only mode available for now\n-    --soloType Droplet\n-\n+    #if str($sc.solo_type) == "CB_UMI_Simple":\n+    @READSHANDLING@\n+    --soloCBwhitelist \'$sc.soloCBwhitelist\'\n     ## 1 - check length of barcode, 0 - do not check\n     ## Good for checking custom chemistries\n-    --soloCBwhitelist \'$soloCBwhitelist\'\n-    --soloBarcodeReadLength $solo.soloBarcodeReadLength\n-\n-    #if str($solo.params.chemistry) == "CR2":\n+    --soloBarcodeReadLength $sc.soloBarcodeReadLength\n+    #if str($sc.params.chemistry) == "CR2":\n     --soloCBstart 1\n     --soloCBlen 16\n     --soloUMIstart 17\n     --soloUMIlen 10\n-    #else if str($solo.params.chemistry) == "CR3":\n+    #else if str($sc.params.chemistry) == "CR3":\n     --soloCBstart 1\n     --soloCBlen 16\n     --soloUMIstart 17\n     --soloUMIlen 12\n-    #else if str($solo.params.chemistry) == "custom":\n-    --soloCBstart $solo.params.soloCBstart\n-    --soloCBlen $solo.params.soloCBlen\n-    --soloUMIstart $solo.params.soloUMIstart\n-    --soloUMIlen $solo.params.soloUMIlen\n+    #else if str($sc.params.chemistry) == "custom":\n+    --soloCBstart $sc.params.soloCBstart\n+    --soloCBlen $sc.params.soloCBlen\n+    --soloUMIstart $sc.params.soloUMIstart\n+    --soloUMIlen $sc.params.soloUMIlen\n     #end if\n \n+    #elif str($sc.solo_type) == "CB_UMI_Complex":\n+    @READSHANDLING@\n+    ## inDrop supports multiple cell barcodes of varying length\n+        #set $cb_whitelist = []\n+        #set $cb_pos = []\n+        #for $cb in $sc.cb_whitelists:\n+            #silent $cb_whitelist.append(str($cb.whitelist_file))\n+            #silent $cb_pos.append(\'_\'.join([str($cb.cb_start_anchor), str($cb.cb_start_anchor_pos),str($cb.cb_end_anchor), str($cb.cb_end_anchor_pos)]))\n+        #end for\n+    #set $cb_whitelist = \' \'.join($cb_whitelist)\n+    --soloCBwhitelist $cb_whitelist\n+    #set $cb_pos = \' \'.join($cb_pos)\n+    --soloCBposition $cb_pos\n+    #set $umi_pos = \'_\'.join([str($sc.umi_start_anchor), str($sc.umi_start_anchor_pos), str($sc.umi_end_anchor), str($sc.umi_end_anchor_pos)])\n+    --soloUMIposition $umi_pos\n+    --soloAdapterSequence $sc.soloAdapterSequence\n+    --soloAdapterMismatchesNmax $sc.soloAdapterMismatchesNmax\n+\n+    #elif str($sc.solo_type) == "SmartSeq":\n+    ## Create a manifest file with fastq files and their corresponding cell-ids\n+    ## For Smart-Seq [R1] is followed by [R2]\n+    --readFilesManifest \'$manifest_file\'\n+        #set $read_files_command = ""\n+        #if str($sc.input_types_smart_seq.use) == "list_single_end":\n+            #if $sc.input_types_smart_seq.single_end_collection[0].is_of_type(\'fastq.gz\', \'fastqsanger.gz\'):\n+                @FASTQ_GZ_OPTION@\n+    '..b'       </element>\n+                            <element name="pair5">\n+                                <collection type="paired">\n+                                    <element name="forward" value="smartseq5.R1.fastq.gz" ftype="fastqsanger.gz" />\n+                                    <element name="reverse" value="smartseq5.R2.fastq.gz" ftype="fastqsanger.gz" />\n+                                </collection>\n+                            </element>\n+                            <element name="pair6">\n+                                <collection type="paired">\n+                                    <element name="forward" value="smartseq6.R1.fastq.gz" ftype="fastqsanger.gz" />\n+                                    <element name="reverse" value="smartseq6.R2.fastq.gz" ftype="fastqsanger.gz" />\n+                                </collection>\n+                            </element>\n+                            <element name="pair7">\n+                                <collection type="paired">\n+                                    <element name="forward" value="smartseq7.R1.fastq.gz" ftype="fastqsanger.gz" />\n+                                    <element name="reverse" value="smartseq7.R2.fastq.gz" ftype="fastqsanger.gz" />\n+                                </collection>\n+                            </element>\n+                            <element name="pair8">\n+                                <collection type="paired">\n+                                    <element name="forward" value="smartseq8.R1.fastq.gz" ftype="fastqsanger.gz" />\n+                                    <element name="reverse" value="smartseq8.R2.fastq.gz" ftype="fastqsanger.gz" />\n+                                </collection>\n+                            </element>\n+                            <element name="pair9">\n+                                <collection type="paired">\n+                                    <element name="forward" value="smartseq9.R1.fastq.gz" ftype="fastqsanger.gz" />\n+                                    <element name="reverse" value="smartseq9.R2.fastq.gz" ftype="fastqsanger.gz" />\n+                                </collection>\n+                            </element>\n+                        </collection>\n+                    </param>\n+                </conditional>\n+                <param name="cell_ids" value="smartseq.cellids.txt" />\n+                <param name="soloUMIdedup" value="Exact" />\n+            </conditional>\n+            <section name="solo" >\n+                <param name="soloStrand" value="Unstranded" />\n+                <conditional name="filter">\n+                    <param name="filter_type" value="topcells" />\n+                    <param name="n_cells" value="2" />\n+                </conditional>\n+            </section>\n+            <output name="output_barcodes_filtered" >\n+                <assert_contents>\n+                    <has_line line="CSC6_D02" />\n+                    <not_has_text text="MGH26_A02" />\n+                </assert_contents>\n+            </output>\n+            <output name="output_genes_filtered">\n+                <assert_contents>\n+                    <has_line_matching expression="ENSG00000279493\\s+FP565260\\.4\\s+Gene\\s+Expression" />\n+                    <has_line_matching expression="ENSG00000279064\\s+FP236315\\.1\\s+Gene\\s+Expression" />\n+                </assert_contents>\n+            </output>\n+            <output name="output_matrix_filtered" >\n+                <assert_contents>\n+                    <has_line_matching expression="14\\s+3\\s+10" />\n+                    <has_line_matching expression="12\\s+3\\s+1" />\n+                </assert_contents>\n+            </output>\n+            <output name="output_stats" >\n+                <assert_contents>\n+                    <has_line_matching expression="\\s+nExactMatch\\s+9000" />\n+                    <has_line_matching expression="\\s+nUMIs\\s+32" />\n+                </assert_contents>\n+            </output>\n         </test>\n     </tests>\n     <help><![CDATA[\n'
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/indrop.R1.fastq.gz
b
Binary file test-data/indrop.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/indrop.R2.fastq.gz
b
Binary file test-data/indrop.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/indrop.barcodes1.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/indrop.barcodes1.txt Fri Jan 15 17:39:11 2021 +0000
b
@@ -0,0 +1,384 @@
+ GTTTGTTT
+ ACCGTGTTT
+ GATAGTGTTT
+ TGAGGCGGTTT
+ GATCGTTT
+ ATCACGTTT
+ GATGTAGTTT
+ TGACACAGTTT
+ CTTTCTTT
+ AGCCTCTTT
+ GACGGGCTTT
+ TGAATGACTTT
+ TGCTATTT
+ ACGGAATTT
+ GACATTTGTT
+ TGAGTTCTGTT
+ CCGCTGTT
+ AAAATCGTT
+ GATTGGCGTT
+ TGACTACCGTT
+ GTAACGTT
+ AACTGAGTT
+ GAAGGCAGTT
+ TGACTGTTCTT
+ ACCTTCTT
+ AATACTCTT
+ GAGAAGGCTT
+ TGAAGGAGCTT
+ TCATCCTT
+ AAGCGCCTT
+ GAGGTCCCTT
+ TGACAATACTT
+ TTGGACTT
+ ACCCGACTT
+ GATCTCACTT
+ TGAGACAACTT
+ TCCTTATT
+ AGATGTATT
+ GAGTCATATT
+ TGAGCCGGATT
+ CTTCGATT
+ AGAACGATT
+ GAACGCCATT
+ TGACATACATT
+ ATCTTTGT
+ ACTACTTGT
+ GAAAGATTGT
+ TGACTTGGTGT
+ TTATCTGT
+ ATGGCCTGT
+ GACGAGATGT
+ TGAGTCCATGT
+ GGGTTGGT
+ ACCCTTGGT
+ GATCTGTGGT
+ TGAAAACTGGT
+ GCATGGGT
+ AAATCGGGT
+ GATTGAGGGT
+ TGATCGACGGT
+ CTTCAGGT
+ AGGGAAGGT
+ GAGAATTCGT
+ TGAGTCGTCGT
+ TTAAGCGT
+ ATGCTCCGT
+ GAACTGCCGT
+ TGATAACCCGT
+ CCAACCGT
+ AGTTTACGT
+ GACAATTAGT
+ TGACGGGTAGT
+ GCTCTAGT
+ AGTATGAGT
+ GATTCCGAGT
+ TGACCAGCAGT
+ TGACCAGT
+ AAGCGAAGT
+ GATGGTTTCT
+ TGACACTTTCT
+ AAGCTTCT
+ ATTGATTCT
+ GATGAGGTCT
+ TGACCTCGTCT
+ GTCTCTCT
+ AGCACCTCT
+ GAGCGTTGCT
+ TGATACGTGCT
+ GGCATGCT
+ AAGATGGCT
+ GAACCACGCT
+ TGAGTGGAGCT
+ TCGAAGCT
+ ATGTGTCCT
+ GACGACTCCT
+ TGATATTGCCT
+ TTCGGCCT
+ AAAACGCCT
+ GACAGTCCCT
+ TGATTTACCCT
+ GCTTACCT
+ AATATACCT
+ GAGGGAACCT
+ TGACCATTACT
+ TAACTACT
+ ATTGTGACT
+ GACACGGACT
+ TGAGAAGCACT
+ GTTCAACT
+ ACCGCAACT
+ GATACAAACT
+ TGACCTGTTAT
+ TAGCTTAT
+ AGGGTGTAT
+ GAGAGAGTAT
+ TGAACATCTAT
+ TTGCATAT
+ AACCCATAT
+ GACGATTGAT
+ TGATCCCTGAT
+ GGTGGGAT
+ AATGCGGAT
+ GAACTAGGAT
+ TGAAGCGCGAT
+ GTTACGAT
+ AGCCAAGAT
+ GAGTTGTCAT
+ TGACAAGTCAT
+ ATATGCAT
+ ACTCCGCAT
+ GAGAGCCCAT
+ TGACAGACCAT
+ CGGCACAT
+ AAAGGTAAT
+ GACGAATAAT
+ TGACTCAGAAT
+ ACTTCAAT
+ AGGGCCAAT
+ GAATGGAAAT
+ TGACAACAAAT
+ AATGTTTG
+ ACTGCGTTG
+ GAATTCCTTG
+ TGAAACCCTTG
+ GTACCTTG
+ ACTAGATTG
+ GAGAGAATTG
+ TGAAGGTTGTG
+ TACTTGTG
+ AGGTTAGTG
+ GAATCAAGTG
+ TGACGAGTCTG
+ CCCATCTG
+ AGCAACCTG
+ GATTAAACTG
+ TGATCGTCATG
+ GCAGCATG
+ AAATGAATG
+ GACCCGAATG
+ TGATAGAAATG
+ AGAGGTGG
+ ACAACGTGG
+ GACTGTCTGG
+ TGATTCGCTGG
+ TCATATGG
+ AGTGGATGG
+ GAGACGATGG
+ TGAATGCATGG
+ CTTACGGG
+ AAGAACGGG
+ GACAAGAGGG
+ TGAAAACAGGG
+ TGCAAGGG
+ AAAAGTCGG
+ GAGATCTCGG
+ TGACGTATCGG
+ ATTTCCGG
+ AAGCTACGG
+ GATAAGACGG
+ TGAAGCGTAGG
+ TAAATAGG
+ ATCATGAGG
+ GATGTAAAGG
+ TGAGACAAAGG
+ GAGTTTCG
+ ATCGGTTCG
+ GACTTCTTCG
+ TGAAAATGTCG
+ TAGCCTCG
+ ATTGGATCG
+ GATGCCATCG
+ TGATTAGTCCG
+ TACAGCCG
+ AACTCACCG
+ GATCGGTACG
+ TGAATTCGACG
+ GTTGCACG
+ AATCCCACG
+ GATGTACACG
+ TGAAACACACG
+ AGGCAACG
+ AACGAAACG
+ GAGGCGTTAG
+ TGATCCCGTAG
+ TAGTCTAG
+ ACGTGCTAG
+ GACCTACTAG
+ TGATGTTTGAG
+ GATGTGAG
+ ATTTGGGAG
+ GATGGAGGAG
+ TGATCACCGAG
+ CTATAGAG
+ AACGCAGAG
+ GACCCTTCAG
+ TGAACGCTCAG
+ CATCGCAG
+ ATCTAGCAG
+ GATGTTCCAG
+ TGAATACCCAG
+ TGCGACAG
+ AGGTCACAG
+ GATTTAACAG
+ TGACACAACAG
+ GGAAACAG
+ AGGCCTAAG
+ GAACACTAAG
+ TGACGTAGAAG
+ GGATAAAG
+ AAGTGAAAG
+ GAGTCCAAAG
+ TGATGTCTTTC
+ CGTATTTC
+ AATATCTTC
+ GATGGGATTC
+ TGAGCGCATTC
+ TTTGTGTC
+ ACAGGTGTC
+ GACGCTAGTC
+ TGAGGTTTCTC
+ TTCCGCTC
+ ACACTCCTC
+ GATGACCCTC
+ TGAGTACACTC
+ TGCGTATC
+ ATCTGCATC
+ GATAACCATC
+ TGAGCCACATC
+ CTTTAATC
+ AAAGTAATC
+ GATCCCAATC
+ TGAGGGAAATC
+ CAGTTTGC
+ ACTGAGTGC
+ GAAGTGATGC
+ TGACTCGATGC
+ GCTTTGGC
+ AATGTTGGC
+ GATACCAGGC
+ TGACACAAGGC
+ ATCAGCGC
+ AGTTACCGC
+ GAGAATACGC
+ TGATTGCACGC
+ AACTTAGC
+ AACGGTAGC
+ GACCCATAGC
+ TGACTACGAGC
+ GGAGAAGC
+ ATTCGTTCC
+ GAGGACTTCC
+ TGATCCAGTCC
+ AGAAGTCC
+ AAAACCTCC
+ GACTTACTCC
+ TGAAACAATCC
+ ACCTTGCC
+ AGAAGTGCC
+ GAATTGGGCC
+ TGATTGTCGCC
+ TTATAGCC
+ AGCAAAGCC
+ GACATCTCCC
+ TGAGTAATCCC
+ TGATGCCC
+ AAATGACCC
+ GACTAGACCC
+ TGAGATTTACC
+ TGGCTACC
+ ATTAGGACC
+ GAGAAAGACC
+ TGATCGACACC
+ GTGTAACC
+ ACCCTAACC
+ GATCTCAACC
+ TGATTGTTTAC
+ CGGCTTAC
+ ACAGATTAC
+ GAAAGCGTAC
+ TGAGTCCGTAC
+ ACGTATAC
+ AGTCAATAC
+ GACTCTTGAC
+ TGAGGTCTGAC
+ AACCTGAC
+ ATAGTGGAC
+ GATGACGGAC
+ TGAGCAAGGAC
+ GATTAGAC
+ ATTCCAGAC
+ GAAGGAAGAC
+ TGAGAGTTCAC
+ TGCCTCAC
+ ATTTATCAC
+ GAATGGGCAC
+ TGACTTCGCAC
+ AGCACCAC
+ AGGTGACAC
+ GACCTGACAC
+ TGACTAGTAAC
+ AGCAGAAC
+ ACGGACAAC
+ GATCGGTTTA
+ TGAAGAAGTTA
+ GGCCCTTA
+ AATGGATTA
+ GACCACATTA
+ TGAGCAGGGTA
+ GAGCGGTA
+ ACTTAGGTA
+ GAGGGAGGTA
+ TGACTCGCGTA
+ CGAACGTA
+ AATTCAGTA
+ GATTGATCTA
+ TGATGTGGCTA
+ ATCCGCTA
+ AAAAGCCTA
+ GACGTACCTA
+ TGAGGCTACTA
+ AGAGACTA
+ ACGTGGATA
+ GAGACAGATA
+ TGATTCACATA
+ CGCTAATA
+ ACCATTTGA
+ GACGCCTTGA
+ TGAGAGGCTGA
+ TGGTATGA
+ AAGCTATGA
+ GATGAAATGA
+ TGACTTCTGGA
+ TCCAGGGA
+ AGTGTCGGA
+ GAACAGCGGA
+ TGAATATAGGA
+ GCAGTCGA
+ AAAACTCGA
+ GAGATTGCGA
+ TGAATGACCGA
+ ACCCACGA
+ AGGGAACGA
+ GAAGTTTAGA
+ TGAGGAATAGA
+ AAATCAGA
+ AGTCAAAGA
+ GACCTATTCA
+ TGAAGGATTCA
+ CGACGTCA
+ ACGCTCTCA
+ GATGTGCTCA
+ TGACTGGTGCA
+ TACCGGCA
+ ATAGTCGCA
+ GACGTCAGCA
+ TGAATGAAGCA
+ CCCAAGCA
+ AGCTTTCCA
+ GATCCGTCCA
+ TGAACTAGCCA
+ AATTCCCA
+ AAGACACCA
+ GAGTTAACCA
+ TGATGATAACA
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/indrop.barcodes2.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/indrop.barcodes2.txt Fri Jan 15 17:39:11 2021 +0000
b
@@ -0,0 +1,384 @@
+ GTTTGTTT
+ CCGTGTTT
+ TAGTGTTT
+ GGCGGTTT
+ GATCGTTT
+ TCACGTTT
+ TGTAGTTT
+ CACAGTTT
+ CTTTCTTT
+ GCCTCTTT
+ CGGGCTTT
+ ATGACTTT
+ TGCTATTT
+ CGGAATTT
+ CATTTGTT
+ GTTCTGTT
+ CCGCTGTT
+ AAATCGTT
+ TTGGCGTT
+ CTACCGTT
+ GTAACGTT
+ ACTGAGTT
+ AGGCAGTT
+ CTGTTCTT
+ ACCTTCTT
+ ATACTCTT
+ GAAGGCTT
+ AGGAGCTT
+ TCATCCTT
+ AGCGCCTT
+ GGTCCCTT
+ CAATACTT
+ TTGGACTT
+ CCCGACTT
+ TCTCACTT
+ GACAACTT
+ TCCTTATT
+ GATGTATT
+ GTCATATT
+ GCCGGATT
+ CTTCGATT
+ GAACGATT
+ ACGCCATT
+ CATACATT
+ ATCTTTGT
+ CTACTTGT
+ AAGATTGT
+ CTTGGTGT
+ TTATCTGT
+ TGGCCTGT
+ CGAGATGT
+ GTCCATGT
+ GGGTTGGT
+ CCCTTGGT
+ TCTGTGGT
+ AAACTGGT
+ GCATGGGT
+ AATCGGGT
+ TTGAGGGT
+ TCGACGGT
+ CTTCAGGT
+ GGGAAGGT
+ GAATTCGT
+ GTCGTCGT
+ TTAAGCGT
+ TGCTCCGT
+ ACTGCCGT
+ TAACCCGT
+ CCAACCGT
+ GTTTACGT
+ CAATTAGT
+ CGGGTAGT
+ GCTCTAGT
+ GTATGAGT
+ TTCCGAGT
+ CCAGCAGT
+ TGACCAGT
+ AGCGAAGT
+ TGGTTTCT
+ CACTTTCT
+ AAGCTTCT
+ TTGATTCT
+ TGAGGTCT
+ CCTCGTCT
+ GTCTCTCT
+ GCACCTCT
+ GCGTTGCT
+ TACGTGCT
+ GGCATGCT
+ AGATGGCT
+ ACCACGCT
+ GTGGAGCT
+ TCGAAGCT
+ TGTGTCCT
+ CGACTCCT
+ TATTGCCT
+ TTCGGCCT
+ AAACGCCT
+ CAGTCCCT
+ TTTACCCT
+ GCTTACCT
+ ATATACCT
+ GGGAACCT
+ CCATTACT
+ TAACTACT
+ TTGTGACT
+ CACGGACT
+ GAAGCACT
+ GTTCAACT
+ CCGCAACT
+ TACAAACT
+ CCTGTTAT
+ TAGCTTAT
+ GGGTGTAT
+ GAGAGTAT
+ ACATCTAT
+ TTGCATAT
+ ACCCATAT
+ CGATTGAT
+ TCCCTGAT
+ GGTGGGAT
+ ATGCGGAT
+ ACTAGGAT
+ AGCGCGAT
+ GTTACGAT
+ GCCAAGAT
+ GTTGTCAT
+ CAAGTCAT
+ ATATGCAT
+ CTCCGCAT
+ GAGCCCAT
+ CAGACCAT
+ CGGCACAT
+ AAGGTAAT
+ CGAATAAT
+ CTCAGAAT
+ ACTTCAAT
+ GGGCCAAT
+ ATGGAAAT
+ CAACAAAT
+ AATGTTTG
+ CTGCGTTG
+ ATTCCTTG
+ AACCCTTG
+ GTACCTTG
+ CTAGATTG
+ GAGAATTG
+ AGGTTGTG
+ TACTTGTG
+ GGTTAGTG
+ ATCAAGTG
+ CGAGTCTG
+ CCCATCTG
+ GCAACCTG
+ TTAAACTG
+ TCGTCATG
+ GCAGCATG
+ AATGAATG
+ CCCGAATG
+ TAGAAATG
+ AGAGGTGG
+ CAACGTGG
+ CTGTCTGG
+ TTCGCTGG
+ TCATATGG
+ GTGGATGG
+ GACGATGG
+ ATGCATGG
+ CTTACGGG
+ AGAACGGG
+ CAAGAGGG
+ AAACAGGG
+ TGCAAGGG
+ AAAGTCGG
+ GATCTCGG
+ CGTATCGG
+ ATTTCCGG
+ AGCTACGG
+ TAAGACGG
+ AGCGTAGG
+ TAAATAGG
+ TCATGAGG
+ TGTAAAGG
+ GACAAAGG
+ GAGTTTCG
+ TCGGTTCG
+ CTTCTTCG
+ AAATGTCG
+ TAGCCTCG
+ TTGGATCG
+ TGCCATCG
+ TTAGTCCG
+ TACAGCCG
+ ACTCACCG
+ TCGGTACG
+ ATTCGACG
+ GTTGCACG
+ ATCCCACG
+ TGTACACG
+ AACACACG
+ AGGCAACG
+ ACGAAACG
+ GGCGTTAG
+ TCCCGTAG
+ TAGTCTAG
+ CGTGCTAG
+ CCTACTAG
+ TGTTTGAG
+ GATGTGAG
+ TTTGGGAG
+ TGGAGGAG
+ TCACCGAG
+ CTATAGAG
+ ACGCAGAG
+ CCCTTCAG
+ ACGCTCAG
+ CATCGCAG
+ TCTAGCAG
+ TGTTCCAG
+ ATACCCAG
+ TGCGACAG
+ GGTCACAG
+ TTTAACAG
+ CACAACAG
+ GGAAACAG
+ GGCCTAAG
+ ACACTAAG
+ CGTAGAAG
+ GGATAAAG
+ AGTGAAAG
+ GTCCAAAG
+ TGTCTTTC
+ CGTATTTC
+ ATATCTTC
+ TGGGATTC
+ GCGCATTC
+ TTTGTGTC
+ CAGGTGTC
+ CGCTAGTC
+ GGTTTCTC
+ TTCCGCTC
+ CACTCCTC
+ TGACCCTC
+ GTACACTC
+ TGCGTATC
+ TCTGCATC
+ TAACCATC
+ GCCACATC
+ CTTTAATC
+ AAGTAATC
+ TCCCAATC
+ GGGAAATC
+ CAGTTTGC
+ CTGAGTGC
+ AGTGATGC
+ CTCGATGC
+ GCTTTGGC
+ ATGTTGGC
+ TACCAGGC
+ CACAAGGC
+ ATCAGCGC
+ GTTACCGC
+ GAATACGC
+ TTGCACGC
+ AACTTAGC
+ ACGGTAGC
+ CCCATAGC
+ CTACGAGC
+ GGAGAAGC
+ TTCGTTCC
+ GGACTTCC
+ TCCAGTCC
+ AGAAGTCC
+ AAACCTCC
+ CTTACTCC
+ AACAATCC
+ ACCTTGCC
+ GAAGTGCC
+ ATTGGGCC
+ TTGTCGCC
+ TTATAGCC
+ GCAAAGCC
+ CATCTCCC
+ GTAATCCC
+ TGATGCCC
+ AATGACCC
+ CTAGACCC
+ GATTTACC
+ TGGCTACC
+ TTAGGACC
+ GAAAGACC
+ TCGACACC
+ GTGTAACC
+ CCCTAACC
+ TCTCAACC
+ TTGTTTAC
+ CGGCTTAC
+ CAGATTAC
+ AAGCGTAC
+ GTCCGTAC
+ ACGTATAC
+ GTCAATAC
+ CTCTTGAC
+ GGTCTGAC
+ AACCTGAC
+ TAGTGGAC
+ TGACGGAC
+ GCAAGGAC
+ GATTAGAC
+ TTCCAGAC
+ AGGAAGAC
+ GAGTTCAC
+ TGCCTCAC
+ TTTATCAC
+ ATGGGCAC
+ CTTCGCAC
+ AGCACCAC
+ GGTGACAC
+ CCTGACAC
+ CTAGTAAC
+ AGCAGAAC
+ CGGACAAC
+ TCGGTTTA
+ AGAAGTTA
+ GGCCCTTA
+ ATGGATTA
+ CCACATTA
+ GCAGGGTA
+ GAGCGGTA
+ CTTAGGTA
+ GGGAGGTA
+ CTCGCGTA
+ CGAACGTA
+ ATTCAGTA
+ TTGATCTA
+ TGTGGCTA
+ ATCCGCTA
+ AAAGCCTA
+ CGTACCTA
+ GGCTACTA
+ AGAGACTA
+ CGTGGATA
+ GACAGATA
+ TTCACATA
+ CGCTAATA
+ CCATTTGA
+ CGCCTTGA
+ GAGGCTGA
+ TGGTATGA
+ AGCTATGA
+ TGAAATGA
+ CTTCTGGA
+ TCCAGGGA
+ GTGTCGGA
+ ACAGCGGA
+ ATATAGGA
+ GCAGTCGA
+ AAACTCGA
+ GATTGCGA
+ ATGACCGA
+ ACCCACGA
+ GGGAACGA
+ AGTTTAGA
+ GGAATAGA
+ AAATCAGA
+ GTCAAAGA
+ CCTATTCA
+ AGGATTCA
+ CGACGTCA
+ CGCTCTCA
+ TGTGCTCA
+ CTGGTGCA
+ TACCGGCA
+ TAGTCGCA
+ CGTCAGCA
+ ATGAAGCA
+ CCCAAGCA
+ GCTTTCCA
+ TCCGTCCA
+ ACTAGCCA
+ AATTCCCA
+ AGACACCA
+ GTTAACCA
+ TGATAACA
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq.cellids.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/test-data/smartseq.cellids.txt Fri Jan 15 17:39:11 2021 +0000
b
@@ -0,0 +1,9 @@
+MGH26_A01
+MGH26_A02
+MGH26_A03
+CSC6_D01
+CSC6_D02
+CSC6_D03
+CSC8_H01
+CSC8_H02
+CSC8_H03
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq1.R1.fastq.gz
b
Binary file test-data/smartseq1.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq1.R2.fastq.gz
b
Binary file test-data/smartseq1.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq2.R1.fastq.gz
b
Binary file test-data/smartseq2.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq2.R2.fastq.gz
b
Binary file test-data/smartseq2.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq3.R1.fastq.gz
b
Binary file test-data/smartseq3.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq3.R2.fastq.gz
b
Binary file test-data/smartseq3.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq4.R1.fastq.gz
b
Binary file test-data/smartseq4.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq4.R2.fastq.gz
b
Binary file test-data/smartseq4.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq5.R1.fastq.gz
b
Binary file test-data/smartseq5.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq5.R2.fastq.gz
b
Binary file test-data/smartseq5.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq6.R1.fastq.gz
b
Binary file test-data/smartseq6.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq6.R2.fastq.gz
b
Binary file test-data/smartseq6.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq7.R1.fastq.gz
b
Binary file test-data/smartseq7.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq7.R2.fastq.gz
b
Binary file test-data/smartseq7.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq8.R1.fastq.gz
b
Binary file test-data/smartseq8.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq8.R2.fastq.gz
b
Binary file test-data/smartseq8.R2.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq9.R1.fastq.gz
b
Binary file test-data/smartseq9.R1.fastq.gz has changed
b
diff -r 00fbfac99d39 -r ec9cbd6b9a49 test-data/smartseq9.R2.fastq.gz
b
Binary file test-data/smartseq9.R2.fastq.gz has changed