Mercurial > repos > iuc > rna_starsolo
comparison rg_rnaStarSolo.xml @ 13:9ee34ba73ebf draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/rgrnastar commit ae6b59a8e52fd34e2347d1fd8d34129c36779266
author | iuc |
---|---|
date | Fri, 17 Feb 2023 20:04:43 +0000 |
parents | 79b885ce78d7 |
children | 1cd2511a396e |
comparison
equal
deleted
inserted
replaced
12:79b885ce78d7 | 13:9ee34ba73ebf |
---|---|
1 <tool id="rna_starsolo" name="RNA STARSolo" version="@VERSION@" profile="20.01" license="MIT"> | 1 <tool id="rna_starsolo" name="RNA STARSolo" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@" license="MIT"> |
2 <description>mapping, demultiplexing and gene quantification for single cell RNA-seq</description> | 2 <description>mapping, demultiplexing and gene quantification for single cell RNA-seq</description> |
3 <macros> | 3 <macros> |
4 <import>macros.xml</import> | 4 <import>macros.xml</import> |
5 </macros> | 5 </macros> |
6 <expand macro="edam"/> | 6 <expand macro="edam"/> |
20 ## Supports Drop-seq, 10X Chromium, inDrop and Smart-Seq | 20 ## Supports Drop-seq, 10X Chromium, inDrop and Smart-Seq |
21 --soloType $sc.solo_type | 21 --soloType $sc.solo_type |
22 | 22 |
23 #if str($sc.solo_type) == "CB_UMI_Simple": | 23 #if str($sc.solo_type) == "CB_UMI_Simple": |
24 @READSHANDLING@ | 24 @READSHANDLING@ |
25 --soloCBwhitelist '$sc.soloCBwhitelist' | 25 #if $sc.soloCBwhitelist: |
26 --soloCBwhitelist '$sc.soloCBwhitelist' | |
27 #else | |
28 --soloCBwhitelist None | |
29 #end if | |
26 ## 1 - check length of barcode, 0 - do not check | 30 ## 1 - check length of barcode, 0 - do not check |
27 ## Good for checking custom chemistries | 31 ## Good for checking custom chemistries |
28 --soloBarcodeReadLength $sc.soloBarcodeReadLength | 32 --soloBarcodeReadLength $sc.soloBarcodeReadLength |
29 #if str($sc.params.chemistry) == "CR2": | 33 #if str($sc.params.chemistry) == "Cv2": |
30 --soloCBstart 1 | 34 --soloCBstart 1 |
31 --soloCBlen 16 | 35 --soloCBlen 16 |
32 --soloUMIstart 17 | 36 --soloUMIstart 17 |
33 --soloUMIlen 10 | 37 --soloUMIlen 10 |
34 #else if str($sc.params.chemistry) == "CR3": | 38 #else if str($sc.params.chemistry) == "Cv3": |
35 --soloCBstart 1 | 39 --soloCBstart 1 |
36 --soloCBlen 16 | 40 --soloCBlen 16 |
37 --soloUMIstart 17 | 41 --soloUMIstart 17 |
38 --soloUMIlen 12 | 42 --soloUMIlen 12 |
39 #else if str($sc.params.chemistry) == "custom": | 43 #else if str($sc.params.chemistry) == "custom": |
88 #end if | 92 #end if |
89 #end if | 93 #end if |
90 --soloCBwhitelist None | 94 --soloCBwhitelist None |
91 #end if | 95 #end if |
92 | 96 |
93 --soloUMIfiltering $solo.soloUMIfiltering | |
94 --soloStrand $solo.soloStrand | 97 --soloStrand $solo.soloStrand |
95 --soloFeatures $solo.soloFeatures | 98 --soloFeatures $solo.soloFeatures |
96 --soloUMIdedup $sc.soloUMIdedup | 99 --soloUMIdedup $sc.umidedup.soloUMIdedup |
97 --quantMode TranscriptomeSAM | 100 #if str($sc.umidedup.soloUMIdedup) == "1MM_CR": |
98 --outSAMtype BAM Unsorted | 101 --soloUMIfiltering $sc.umidedup.soloUMIfiltering |
102 #end if | |
103 --quantMode TranscriptomeSAM $solo.quantModeGene | |
104 #set $tag_names = str($solo.outSAMattributes).replace(',', ' ') | |
105 --outSAMattributes $tag_names | |
106 #if "CB" in $tag_names or "UB" in $tag_names or str($outWig.outWigType) != 'None': | |
107 --outSAMtype BAM SortedByCoordinate | |
108 #else: | |
109 --outSAMtype BAM Unsorted | |
110 #end if | |
99 | 111 |
100 #if str($solo.filter.filter_type) == "cellranger2": | 112 #if str($solo.filter.filter_type) == "cellranger2": |
101 --soloCellFilter CellRanger2.2 $solo.filter.n_expected $solo.filter.max_perc $solo.filter.max_min_ratio | 113 --soloCellFilter CellRanger2.2 $solo.filter.n_expected $solo.filter.max_perc $solo.filter.max_min_ratio |
102 #else if str($solo.filter.filter_type) == "emptydrops": | 114 #else if str($solo.filter.filter_type) == "emptydrops": |
103 --soloCellFilter EmptyDrops_CR $solo.filter.nExpectedCells $solo.filter.maxPercentile $solo.filter.maxMinRatio $solo.filter.indMin $solo.filter.indMax $solo.filter.umiMin $solo.filter.umiMinFracMedian $solo.filter.candMaxN $solo.filter.FDR $solo.filter.simN | 115 --soloCellFilter EmptyDrops_CR $solo.filter.nExpectedCells $solo.filter.maxPercentile $solo.filter.maxMinRatio $solo.filter.indMin $solo.filter.indMax $solo.filter.umiMin $solo.filter.umiMinFracMedian $solo.filter.candMaxN $solo.filter.FDR $solo.filter.simN |
107 --soloCellFilter None | 119 --soloCellFilter None |
108 #end if | 120 #end if |
109 ## Splice junctions are always under "raw" directory | 121 ## Splice junctions are always under "raw" directory |
110 | 122 |
111 --soloOutFormatFeaturesGeneField3 '${solo.soloOutFormatFeaturesGeneField3}' | 123 --soloOutFormatFeaturesGeneField3 '${solo.soloOutFormatFeaturesGeneField3}' |
124 | |
125 ## Limits | |
126 @LIMITS@ | |
127 | |
128 ##outWig: | |
129 @OUTWIG@ | |
112 ## Rename the the selected features directory | 130 ## Rename the the selected features directory |
113 && mv Solo.out/${solo.soloFeatures} Solo.out/soloFeatures | 131 && mv Solo.out/${solo.soloFeatures} Solo.out/soloFeatures |
114 ## put the barcodes and features stats into a single file | 132 ## put the barcodes and features stats into a single file |
115 && cat <(echo "Barcodes:") Solo.out/Barcodes.stats <(echo "Genes:") Solo.out/soloFeatures/Features.stats > '${output_stats}' | 133 && cat <(echo "Barcodes:") Solo.out/Barcodes.stats <(echo "Genes:") Solo.out/soloFeatures/Features.stats > '${output_stats}' |
116 | 134 |
117 ## BAM sorting (logic copied from samtools_sort wrapper) | 135 |
118 ## choosing BAM SortedByCoord appeared once to give fewer reads | 136 #if "CB" in $tag_names or "UB" in $tag_names or str($outWig.outWigType) != 'None': |
119 ## than BAM Unsorted followed by a samtools sort | 137 ## recompress BAM output for smaller file size |
120 ## so better go with the latter? | 138 && samtools view -b -o '$output_BAM' Aligned.sortedByCoord.out.bam |
121 | 139 #else: |
122 && | 140 ## BAM sorting (logic copied from samtools_sort wrapper) |
123 ##compute the number of ADDITIONAL threads to be used by samtools (-@) | 141 ## choosing BAM SortedByCoord appeared once to give fewer reads |
124 addthreads=\${GALAXY_SLOTS:-2} && (( addthreads-- )) && | 142 ## than BAM Unsorted followed by a samtools sort |
125 ##compute the number of memory available to samtools sort (-m) | 143 ## so better go with the latter? |
126 ##use only 75% of available: https://github.com/samtools/samtools/issues/831 | 144 |
127 addmemory=\${GALAXY_MEMORY_MB_PER_SLOT:-768} && | 145 && |
128 ((addmemory=addmemory*75/100)) && | 146 ##compute the number of ADDITIONAL threads to be used by samtools (-@) |
129 samtools sort -@ \$addthreads -m \$addmemory"M" -T "\${TMPDIR:-.}" -O bam -o '$output_BAM' Aligned.out.bam | 147 addthreads=\${GALAXY_SLOTS:-2} && (( addthreads-- )) && |
148 ##compute the number of memory available to samtools sort (-m) | |
149 ##use only 75% of available: https://github.com/samtools/samtools/issues/831 | |
150 addmemory=\${GALAXY_MEMORY_MB_PER_SLOT:-768} && | |
151 ((addmemory=addmemory*75/100)) && | |
152 samtools sort -@ \$addthreads -m \$addmemory"M" -T "\${TMPDIR:-.}" -O bam -o '$output_BAM' Aligned.out.bam | |
153 #end if | |
154 ##outWig: | |
155 @OUTWIGOUTPUTS@ | |
156 | |
130 ]]></command> | 157 ]]></command> |
131 <configfiles> | 158 <configfiles> |
132 <configfile name="manifest_file" > | 159 <configfile name="manifest_file" > |
133 #if str($sc.solo_type) == "SmartSeq": | 160 #if str($sc.solo_type) == "SmartSeq": |
134 #set $cellids_fh = open(str($sc.cell_ids), 'r') | 161 #set $cellids_fh = open(str($sc.cell_ids), 'r') |
168 <when value="with-gtf"> | 195 <when value="with-gtf"> |
169 <expand macro="index_selection" with_gene_model="1" /> | 196 <expand macro="index_selection" with_gene_model="1" /> |
170 </when> | 197 </when> |
171 <when value="without-gtf"> | 198 <when value="without-gtf"> |
172 <expand macro="index_selection" with_gene_model="0" /> | 199 <expand macro="index_selection" with_gene_model="0" /> |
173 <expand macro="@SJDBOPTIONS@" optional="false" /> | 200 <expand macro="SJDBOPTIONS"/> |
174 </when> | 201 </when> |
175 </conditional> | 202 </conditional> |
176 </when> | 203 </when> |
177 <when value="history"> | 204 <when value="history"> |
178 <expand macro="ref_selection" /> | 205 <expand macro="ref_selection" /> |
179 <expand macro="@SJDBOPTIONS@" optional="false"/> | 206 <expand macro="SJDBOPTIONS"/> |
180 </when> | 207 </when> |
181 </conditional> | 208 </conditional> |
182 <conditional name="sc" > | 209 <conditional name="sc" > |
183 <param name="solo_type" type="select" label="Type of single-cell RNA-seq" > | 210 <param name="solo_type" type="select" label="Type of single-cell RNA-seq" > |
184 <option value="CB_UMI_Simple">Drop-seq or 10X Chromium</option> | 211 <option value="CB_UMI_Simple">Drop-seq or 10X Chromium</option> |
185 <option value="CB_UMI_Complex">inDrop</option> | 212 <option value="CB_UMI_Complex">inDrop</option> |
186 <option value="SmartSeq">Smart-Seq</option> | 213 <option value="SmartSeq">Smart-Seq</option> |
187 </param> | 214 </param> |
188 <when value="CB_UMI_Simple"> | 215 <when value="CB_UMI_Simple"> |
189 <expand macro="input_selection" /> | 216 <expand macro="input_selection" /> |
190 <param format="txt,tsv" argument="--soloCBwhitelist" type="data" label="RNA-Seq Cell Barcode Whitelist"/> | 217 <param format="txt,tsv" argument="--soloCBwhitelist" optional="True" type="data" label="RNA-Seq Cell Barcode Whitelist"/> |
191 <conditional name="params" > | 218 <conditional name="params" > |
192 <param name="chemistry" type="select" label="Configure Chemistry Options"> | 219 <param name="chemistry" type="select" label="Configure Chemistry Options"> |
193 <option value="CR2" selected="true">Cell Ranger v2</option> | 220 <option value="Cv2" selected="true">Chromium chemistry v2</option> |
194 <option value="CR3">Cell Ranger v3</option> | 221 <option value="Cv3">Chromium chemistry v3</option> |
195 <option value="custom">Custom</option> | 222 <option value="custom">Custom</option> |
196 </param> | 223 </param> |
197 <when value="CR2" /> | 224 <when value="Cv2" /> |
198 <when value="CR3" /> | 225 <when value="Cv3" /> |
199 <when value="custom" > | 226 <when value="custom" > |
200 <param argument="--soloCBstart" type="integer" min="1" value="1" label="Cell Barcode Start Base" /> | 227 <param argument="--soloCBstart" type="integer" min="1" value="1" label="Cell Barcode Start Base" /> |
201 <param argument="--soloCBlen" type="integer" min="1" value="16" label="Cell Barcode Length" /> | 228 <param argument="--soloCBlen" type="integer" min="1" value="16" label="Cell Barcode Length" /> |
202 <param argument="--soloUMIstart" type="integer" min="1" value="17" label="UMI Start Base" /> | 229 <param argument="--soloUMIstart" type="integer" min="1" value="17" label="UMI Start Base" /> |
203 <param argument="--soloUMIlen" type="integer" min="1" value="10" label="UMI Length" /> | 230 <param argument="--soloUMIlen" type="integer" min="1" value="10" label="UMI Length" /> |
217 </conditional> | 244 </conditional> |
218 <expand macro="solo_adapter_params" /> | 245 <expand macro="solo_adapter_params" /> |
219 </when> | 246 </when> |
220 </conditional> | 247 </conditional> |
221 <param argument="--soloBarcodeReadLength" type="boolean" truevalue="1" falsevalue="0" checked="true" label="Barcode Size is same size of the Read" help="Disable this if your R1 barcodes contain poly-T bases after the barcode sequence." /> | 248 <param argument="--soloBarcodeReadLength" type="boolean" truevalue="1" falsevalue="0" checked="true" label="Barcode Size is same size of the Read" help="Disable this if your R1 barcodes contain poly-T bases after the barcode sequence." /> |
222 <param argument="--soloUMIdedup" type="select" label="UMI deduplication (collapsing) algorithm" help="All has all UMIs with 1 mismatch distance to each other collapsed, Directional follows the 'directional' method given in UMI-tools, Exact collapses only exactly matching UMIs."> | 249 <conditional name="umidedup"> |
223 <expand macro="umidedup_options" /> | 250 <param argument="--soloUMIdedup" type="select" label="UMI deduplication (collapsing) algorithm" help="All has all UMIs with 1 mismatch distance to each other collapsed, Directional follows the 'directional' method given in UMI-tools, Exact collapses only exactly matching UMIs."> |
224 <option value="Exact" >Exact</option> | 251 <expand macro="umidedup_options" /> |
225 <option value="1MM_CR" >CellRanger2-4 algorithm</option> | 252 <option value="Exact" >Exact</option> |
226 </param> | 253 <option value="1MM_CR" >CellRanger2-4 algorithm</option> |
254 </param> | |
255 <when value="1MM_All"/> | |
256 <when value="1MM_Directional_UMItools"/> | |
257 <when value="1MM_Directional"/> | |
258 <when value="Exact"/> | |
259 <when value="1MM_CR"> | |
260 <param argument="--soloUMIfiltering" type="select" label="Type of UMI filtering" > | |
261 <option value="-" selected="true">Remove UMIs with N and homopolymers (similar to CellRanger 2.2.0)</option> | |
262 <option value="MultiGeneUMI" >Remove lower-count UMIs that map to more than one gene</option> | |
263 <option value="MultiGeneUMI_All" >Remove all UMIs that map to more than one gene</option> | |
264 <option value="MultiGeneUMI_CR" >Remove lower-count UMIs that map to more than one gene, matching CellRanger > 3.0.0</option> | |
265 </param> | |
266 </when> | |
267 </conditional> | |
227 <param argument="--soloCBmatchWLtype" type="select" label="Matching the Cell Barcodes to the WhiteList" help="Exact: only exact matches allowed; 1MM: only one match in whitelist with 1 mismatched base allowed. Allowed | 268 <param argument="--soloCBmatchWLtype" type="select" label="Matching the Cell Barcodes to the WhiteList" help="Exact: only exact matches allowed; 1MM: only one match in whitelist with 1 mismatched base allowed. Allowed |
228 CBs have to have at least one read with exact match; 1MM_multi: multiple matches in whitelist with 1 mismatched base allowed, posterior probability calculation is used choose one of the matches; 1MM_multi_pseudocounts: same as 1MM_Multi, but pseudocounts of 1 are added to all whitelist barcodes."> | 269 CBs have to have at least one read with exact match; 1MM_multi: multiple matches in whitelist with 1 mismatched base allowed, posterior probability calculation is used choose one of the matches; 1MM_multi_pseudocounts: same as 1MM_Multi, but pseudocounts of 1 are added to all whitelist barcodes."> |
229 <expand macro="cb_match_wl_common" /> | 270 <expand macro="cb_match_wl_common" /> |
230 <expand macro="cb_match_wl_cellranger" /> | 271 <expand macro="cb_match_wl_cellranger" /> |
231 </param> | 272 </param> |
250 <param name="umi_end_anchor" type="select" label="End anchor base for UMI"> | 291 <param name="umi_end_anchor" type="select" label="End anchor base for UMI"> |
251 <expand macro="anchor_types" /> | 292 <expand macro="anchor_types" /> |
252 </param> | 293 </param> |
253 <param name="umi_end_anchor_pos" type="integer" value="0" label="0-based position of the UMI end with respect to the anchor base" /> | 294 <param name="umi_end_anchor_pos" type="integer" value="0" label="0-based position of the UMI end with respect to the anchor base" /> |
254 <expand macro="solo_adapter_params" /> | 295 <expand macro="solo_adapter_params" /> |
255 <param argument="--soloUMIdedup" type="select" label="UMI deduplication (collapsing) algorithm" help="All has all UMIs with 1 mismatch distance to each other collapsed, Directional follows the 'directional' method given in UMI-tools, Exact collapses only exactly matching UMIs."> | 296 <conditional name="umidedup"> |
256 <expand macro="umidedup_options" /> | 297 <param argument="--soloUMIdedup" type="select" label="UMI deduplication (collapsing) algorithm" help="All has all UMIs with 1 mismatch distance to each other collapsed, Directional follows the 'directional' method given in UMI-tools, Exact collapses only exactly matching UMIs."> |
257 <option value="Exact" >Exact</option> | 298 <expand macro="umidedup_options" /> |
258 <option value="1MM_CR" >CellRanger2-4 algorithm</option> | 299 <option value="Exact" >Exact</option> |
259 </param> | 300 <option value="1MM_CR" >CellRanger2-4 algorithm</option> |
301 </param> | |
302 <when value="1MM_All"/> | |
303 <when value="1MM_Directional_UMItools"/> | |
304 <when value="1MM_Directional"/> | |
305 <when value="Exact"/> | |
306 <when value="1MM_CR"> | |
307 <param argument="--soloUMIfiltering" type="select" label="Type of UMI filtering" > | |
308 <option value="-" selected="true">Remove UMIs with N and homopolymers (similar to CellRanger 2.2.0)</option> | |
309 <option value="MultiGeneUMI" >Remove lower-count UMIs that map to more than one gene</option> | |
310 <option value="MultiGeneUMI_All" >Remove all UMIs that map to more than one gene</option> | |
311 <option value="MultiGeneUMI_CR" >Remove lower-count UMIs that map to more than one gene, matching CellRanger > 3.0.0</option> | |
312 </param> | |
313 </when> | |
314 </conditional> | |
260 <param argument="--soloCBmatchWLtype" type="select" label="Matching the Cell Barcodes to the WhiteList" help="Exact: only exact matches allowed; 1MM: only one match in whitelist with 1 mismatched base allowed. Allowed | 315 <param argument="--soloCBmatchWLtype" type="select" label="Matching the Cell Barcodes to the WhiteList" help="Exact: only exact matches allowed; 1MM: only one match in whitelist with 1 mismatched base allowed. Allowed |
261 CBs have to have at least one read with exact match; 1MM_multi: multiple matches in whitelist with 1 mismatched base allowed, posterior probability calculation is used choose one of the matches; 1MM_multi_pseudocounts: same as 1MM_Multi, but pseudocounts of 1 are added to all whitelist barcodes."> | 316 CBs have to have at least one read with exact match; 1MM_multi: multiple matches in whitelist with 1 mismatched base allowed, posterior probability calculation is used choose one of the matches; 1MM_multi_pseudocounts: same as 1MM_Multi, but pseudocounts of 1 are added to all whitelist barcodes."> |
262 <expand macro="cb_match_wl_common" /> | 317 <expand macro="cb_match_wl_common" /> |
318 <!-- should we add EditDist_2? --> | |
263 </param> | 319 </param> |
264 </when> | 320 </when> |
265 <when value="SmartSeq"> | 321 <when value="SmartSeq"> |
266 <expand macro="input_selection_smart_seq" /> | 322 <expand macro="input_selection_smart_seq" /> |
267 <param name="cell_ids" format="txt,tsv" type="data" label="File containing cell IDs of the samples. One ID per line in order of samples in the above collection."/> | 323 <param name="cell_ids" format="txt,tsv" type="data" label="File containing cell IDs of the samples. One ID per line in order of samples in the above collection."/> |
268 <param argument="--soloUMIdedup" type="select" label="UMI deduplication (collapsing) algorithm" help="All has all UMIs with 1 mismatch distance to each other collapsed, Directional follows the 'directional' method given in UMI-tools, Exact collapses only exactly matching UMIs."> | 324 <conditional name="umidedup"> |
269 <option value="Exact" >Exact</option> | 325 <param argument="--soloUMIdedup" type="select" label="UMI deduplication (collapsing) algorithm" help="All has all UMIs with 1 mismatch distance to each other collapsed, Directional follows the 'directional' method given in UMI-tools, Exact collapses only exactly matching UMIs."> |
270 <option value="NoDedup">Do not deduplicate UMIs</option> | 326 <expand macro="umidedup_options" /> |
271 </param> | 327 <option value="Exact" >Exact</option> |
328 <option value="NoDedup" >CellRanger2-4 algorithm</option> | |
329 </param> | |
330 <when value="1MM_All"/> | |
331 <when value="1MM_Directional_UMItools"/> | |
332 <when value="1MM_Directional"/> | |
333 <when value="Exact"/> | |
334 <when value="NoDedup"/> | |
335 </conditional> | |
272 </when> | 336 </when> |
273 </conditional> | 337 </conditional> |
274 <section name="solo" title="Advanced Settings" expanded="true"> | 338 <section name="solo" title="Advanced Settings" expanded="true"> |
275 <param argument="--soloStrand" type="select" label="Strandedness of Library" help="Unstranded has no strand information, Forward has the read strand the same as the original RNA molecule, Reverse has the read strand opposite to the original RNA molecule"> | 339 <param argument="--soloStrand" type="select" label="Strandedness of Library" help="Unstranded has no strand information, Forward has the read strand the same as the original RNA molecule, Reverse has the read strand opposite to the original RNA molecule"> |
276 <option value="Unstranded" >No strand information</option> | 340 <option value="Unstranded" >No strand information</option> |
279 </param> | 343 </param> |
280 <param argument="--soloFeatures" type="select" label="Collect UMI counts for these genomic features" > | 344 <param argument="--soloFeatures" type="select" label="Collect UMI counts for these genomic features" > |
281 <option value="Gene" selected="true">Gene: Count reads matching the Gene Transcript</option> | 345 <option value="Gene" selected="true">Gene: Count reads matching the Gene Transcript</option> |
282 <option value="SJ" >Splice Junctions: Count reads at exon-intron junctions</option> | 346 <option value="SJ" >Splice Junctions: Count reads at exon-intron junctions</option> |
283 <option value="GeneFull" >Full: Count all reads overlapping genes' exons and introns</option> | 347 <option value="GeneFull" >Full: Count all reads overlapping genes' exons and introns</option> |
284 </param> | 348 <option value="GeneFull_ExonOverIntron" >Full: Count all reads overlapping genes' exons and introns: prioritize 100% overlap with exons</option> |
285 <param argument="--soloUMIfiltering" type="select" label="Type of UMI filtering" > | 349 <option value="GeneFull_Ex50pAS" >Full: Count all reads overlapping genes' exons and introns: prioritize 50% overlap with exons. Do not count reads with 100% exonic overlap in the antisense direction.</option> |
286 <option value="-" selected="true">Remove UMIs with N and homopolymers (similar to CellRanger 2.2.0)</option> | |
287 <option value="MultiGeneUMI" >Remove lower-count UMIs that map to more than one gene</option> | |
288 <option value="MultiGeneUMI_CR" >Remove lower-count UMIs that map to more than one gene, matching CellRanger > 3.0.0</option> | |
289 </param> | 350 </param> |
290 <conditional name="filter" > | 351 <conditional name="filter" > |
291 <param name="filter_type" type="select" label="Cell filtering type and parameters" > | 352 <param name="filter_type" type="select" label="Cell filtering type and parameters" > |
292 <option value="cellranger2" selected="true" >Simple filtering of CellRanger v2</option> | 353 <option value="cellranger2" selected="true" >Simple filtering of CellRanger v2</option> |
293 <option value="emptydrops" >EmptyDrops filtering in CellRanger flavor</option> | 354 <option value="emptydrops" >EmptyDrops filtering in CellRanger flavor</option> |
296 </param> | 357 </param> |
297 <when value="cellranger2" > | 358 <when value="cellranger2" > |
298 <param name="n_expected" type="integer" min="1" value="3000" label="Number of expected cells" /> | 359 <param name="n_expected" type="integer" min="1" value="3000" label="Number of expected cells" /> |
299 <param name="max_perc" type="float" min="0" max="1" value="0.99" label="Robust maximum percentile for UMI count" /> | 360 <param name="max_perc" type="float" min="0" max="1" value="0.99" label="Robust maximum percentile for UMI count" /> |
300 <param name="max_min_ratio" type="float" min="1" value="10" label="Maximum to minimum ratio for UMI count" /> | 361 <param name="max_min_ratio" type="float" min="1" value="10" label="Maximum to minimum ratio for UMI count" /> |
362 <param name="output_raw" type="boolean" checked="false" label="Output raw matrix in addition to filtered one" /> | |
301 </when> | 363 </when> |
302 <when value="emptydrops" > | 364 <when value="emptydrops" > |
303 <param name="nExpectedCells" type="integer" min="1" value="3000" label="Number of expected cells" /> | 365 <param name="nExpectedCells" type="integer" min="1" value="3000" label="Number of expected cells" /> |
304 <param name="maxPercentile" type="float" min="0" max="1" value="0.99" label="Robust maximum percentile for UMI count" /> | 366 <param name="maxPercentile" type="float" min="0" max="1" value="0.99" label="Robust maximum percentile for UMI count" /> |
305 <param name="maxMinRatio" type="float" min="1" value="10" label="Maximum to minimum ratio for UMI count" /> | 367 <param name="maxMinRatio" type="float" min="1" value="10" label="Maximum to minimum ratio for UMI count" /> |
308 <param name="umiMin" type="integer" value="500" label="Consider at least these many UMIs per barcode after initial cell calling" /> | 370 <param name="umiMin" type="integer" value="500" label="Consider at least these many UMIs per barcode after initial cell calling" /> |
309 <param name="umiMinFracMedian" type="float" value="0.01" label="Minimum UMI:median ratio after initial cell calling" /> | 371 <param name="umiMinFracMedian" type="float" value="0.01" label="Minimum UMI:median ratio after initial cell calling" /> |
310 <param name="candMaxN" type="integer" value="20000" label="Number of extra barcodes after initial cell calling" /> | 372 <param name="candMaxN" type="integer" value="20000" label="Number of extra barcodes after initial cell calling" /> |
311 <param name="FDR" type="float" value="0.01" label="Maximum adjusted p-value for determining a barcode as non-ambient" /> | 373 <param name="FDR" type="float" value="0.01" label="Maximum adjusted p-value for determining a barcode as non-ambient" /> |
312 <param name="simN" type="integer" value="10000" label="Number of log likelihood simulations" /> | 374 <param name="simN" type="integer" value="10000" label="Number of log likelihood simulations" /> |
375 <param name="output_raw" type="boolean" checked="false" label="Output raw matrix in addition to filtered one" /> | |
313 </when> | 376 </when> |
314 <when value="topcells" > | 377 <when value="topcells" > |
315 <param name="n_cells" type="integer" min="1" value="3000" label="Number of top cells to report sorted by UMI count" /> | 378 <param name="n_cells" type="integer" min="1" value="3000" label="Number of top cells to report sorted by UMI count" /> |
379 <param name="output_raw" type="boolean" checked="false" label="Output raw matrix in addition to filtered one" /> | |
316 </when> | 380 </when> |
317 <when value="no_filter" /> | 381 <when value="no_filter"> |
382 <param name="output_raw" type="hidden" value="true" /> | |
383 </when> | |
318 </conditional> | 384 </conditional> |
319 <param argument="--soloOutFormatFeaturesGeneField3" type="text" value="Gene Expression" label="Field 3 in the Genes output." help="Input '-' to remove the 3rd column from the output." /> | 385 <param argument="--soloOutFormatFeaturesGeneField3" type="text" value="Gene Expression" label="Field 3 in the Genes output." help="Input '-' to remove the 3rd column from the output." /> |
386 <param argument="--outSAMattributes" type="select" display="checkboxes" multiple="true" optional="true" | |
387 label="Read alignment tags to include in the BAM output"> | |
388 <expand macro="common_SAM_attributes"/> | |
389 <option value="CR">CR Cellular barcode sequence bases (uncorrected)</option> | |
390 <option value="CY">CY Phred quality of the cellular barcode sequence in the CR tag</option> | |
391 <option value="GX">GX Gene ID</option> | |
392 <option value="GN">GN Gene name</option> | |
393 <option value="CB">CB Cell identifier (corrected)</option> | |
394 <option value="UB">UB UMI (corrected)</option> | |
395 </param> | |
396 <param name="quantModeGene" type="boolean" truevalue="GeneCounts" falsevalue="" checked="false" label="Output global gene count" help="Can be used by MultiQC" /> | |
397 <expand macro="limits" /> | |
320 </section> | 398 </section> |
399 <expand macro="outWig"/> | |
321 </inputs> | 400 </inputs> |
322 <outputs> | 401 <outputs> |
323 <data format="txt" name="output_log" label="${tool.name} on ${on_string}: log" from_work_dir="Log.final.out"> | 402 <data format="txt" name="output_log" label="${tool.name} on ${on_string}: log" from_work_dir="Log.final.out"> |
324 <expand macro="dbKeyActions" /> | 403 <expand macro="dbKeyActions" /> |
325 </data> | 404 </data> |
331 </data> | 410 </data> |
332 --> | 411 --> |
333 <!-- soloCellFilter set to None, if SJ is selected for soloFeatures --> | 412 <!-- soloCellFilter set to None, if SJ is selected for soloFeatures --> |
334 <data format="tsv" name="output_genes" label="${tool.name} on ${on_string}: Genes raw" | 413 <data format="tsv" name="output_genes" label="${tool.name} on ${on_string}: Genes raw" |
335 from_work_dir="Solo.out/soloFeatures/raw/features.tsv" > | 414 from_work_dir="Solo.out/soloFeatures/raw/features.tsv" > |
336 <filter>solo['filter']['filter_type'] == "no_filter" or solo['soloFeatures'] == "SJ" </filter> | 415 <filter>solo['filter']['output_raw'] or solo['soloFeatures'] == "SJ" </filter> |
337 </data> | 416 </data> |
338 <data format="tsv" name="output_genes_filtered" label="${tool.name} on ${on_string}: Genes filtered" | 417 <data format="tsv" name="output_genes_filtered" label="${tool.name} on ${on_string}: Genes filtered" |
339 from_work_dir="Solo.out/soloFeatures/filtered/features.tsv" > | 418 from_work_dir="Solo.out/soloFeatures/filtered/features.tsv" > |
340 <filter>solo['filter']['filter_type'] != "no_filter" and solo['soloFeatures'] != "SJ" </filter> | 419 <filter>solo['filter']['filter_type'] != "no_filter" and solo['soloFeatures'] != "SJ" </filter> |
341 </data> | 420 </data> |
342 <data format="tsv" name="output_barcodes" label="${tool.name} on ${on_string}: Barcodes raw" | 421 <data format="tsv" name="output_barcodes" label="${tool.name} on ${on_string}: Barcodes raw" |
343 from_work_dir="Solo.out/soloFeatures/raw/barcodes.tsv" > | 422 from_work_dir="Solo.out/soloFeatures/raw/barcodes.tsv" > |
344 <filter>solo['filter']['filter_type'] == "no_filter" or solo['soloFeatures'] == "SJ" </filter> | 423 <filter>solo['filter']['output_raw'] or solo['soloFeatures'] == "SJ" </filter> |
345 </data> | 424 </data> |
346 <data format="tsv" name="output_barcodes_filtered" label="${tool.name} on ${on_string}: Barcodes filtered" | 425 <data format="tsv" name="output_barcodes_filtered" label="${tool.name} on ${on_string}: Barcodes filtered" |
347 from_work_dir="Solo.out/soloFeatures/filtered/barcodes.tsv" > | 426 from_work_dir="Solo.out/soloFeatures/filtered/barcodes.tsv" > |
348 <filter>solo['filter']['filter_type'] != "no_filter" and solo['soloFeatures'] != "SJ" </filter> | 427 <filter>solo['filter']['filter_type'] != "no_filter" and solo['soloFeatures'] != "SJ" </filter> |
349 </data> | 428 </data> |
350 <data format="mtx" name="output_matrix" label="${tool.name} on ${on_string}: Matrix Gene Counts raw" | 429 <data format="mtx" name="output_matrix" label="${tool.name} on ${on_string}: Matrix Gene Counts raw" |
351 from_work_dir="Solo.out/soloFeatures/raw/matrix.mtx" > | 430 from_work_dir="Solo.out/soloFeatures/raw/matrix.mtx" > |
352 <filter>solo['soloFeatures'] == "Gene" and solo['filter']['filter_type'] == "no_filter" </filter> | 431 <filter>solo['soloFeatures'] == "Gene" and solo['filter']['output_raw'] </filter> |
353 <expand macro="dbKeyActions" /> | 432 <expand macro="dbKeyActions" /> |
354 </data> | 433 </data> |
355 <data format="mtx" name="output_matrix_filtered" label="${tool.name} on ${on_string}: Matrix Gene Counts filtered" | 434 <data format="mtx" name="output_matrix_filtered" label="${tool.name} on ${on_string}: Matrix Gene Counts filtered" |
356 from_work_dir="Solo.out/soloFeatures/filtered/matrix.mtx" > | 435 from_work_dir="Solo.out/soloFeatures/filtered/matrix.mtx" > |
357 <filter>solo['soloFeatures'] == "Gene" and solo['filter']['filter_type'] != "no_filter" </filter> | 436 <filter>solo['soloFeatures'] == "Gene" and solo['filter']['filter_type'] != "no_filter" </filter> |
362 <filter>solo['soloFeatures'] == "SJ" </filter> | 441 <filter>solo['soloFeatures'] == "SJ" </filter> |
363 <expand macro="dbKeyActions" /> | 442 <expand macro="dbKeyActions" /> |
364 </data> | 443 </data> |
365 <data format="mtx" name="output_matrixGeneFull" label="${tool.name} on ${on_string}: Matrix Full Gene Counts raw" | 444 <data format="mtx" name="output_matrixGeneFull" label="${tool.name} on ${on_string}: Matrix Full Gene Counts raw" |
366 from_work_dir="Solo.out/soloFeatures/raw/matrix.mtx" > | 445 from_work_dir="Solo.out/soloFeatures/raw/matrix.mtx" > |
367 <filter>solo['soloFeatures'] == "GeneFull" and solo['filter']['filter_type'] == "no_filter" </filter> | 446 <filter>"GeneFull" in solo['soloFeatures'] and solo['filter']['output_raw'] </filter> |
368 <expand macro="dbKeyActions" /> | 447 <expand macro="dbKeyActions" /> |
369 </data> | 448 </data> |
370 <data format="mtx" name="output_matrixGeneFull_filtered" label="${tool.name} on ${on_string}: Matrix Full Gene Counts filtered" | 449 <data format="mtx" name="output_matrixGeneFull_filtered" label="${tool.name} on ${on_string}: Matrix Full Gene Counts filtered" |
371 from_work_dir="Solo.out/soloFeatures/filtered/matrix.mtx" > | 450 from_work_dir="Solo.out/soloFeatures/filtered/matrix.mtx" > |
372 <filter>solo['soloFeatures'] == "GeneFull" and solo['filter']['filter_type'] != "no_filter" </filter> | 451 <filter>"GeneFull" in solo['soloFeatures'] and solo['filter']['filter_type'] != "no_filter" </filter> |
373 <expand macro="dbKeyActions" /> | 452 <expand macro="dbKeyActions" /> |
374 </data> | 453 </data> |
375 <data format="bam" name="output_BAM" label="${tool.name} on ${on_string}: Alignments" > | 454 <data format="bam" name="output_BAM" label="${tool.name} on ${on_string}: Alignments" > |
376 <expand macro="dbKeyActions" /> | 455 <expand macro="dbKeyActions" /> |
377 </data> | 456 </data> |
378 <data format="txt" name="output_stats" label="${tool.name} on ${on_string}: Barcode/Feature Statistic Summaries"/> | 457 <data format="txt" name="output_stats" label="${tool.name} on ${on_string}: Barcode/Feature Statistic Summaries"/> |
458 <data name="reads_per_gene" format="tabular" label="${tool.name} on ${on_string}: combined reads per gene" from_work_dir="ReadsPerGene.out.tab"> | |
459 <filter>solo['quantModeGene']</filter> | |
460 <expand macro="dbKeyActions" /> | |
461 <expand macro="outCountActions" /> | |
462 </data> | |
463 <expand macro="outWigOutputs"/> | |
379 </outputs> | 464 </outputs> |
380 <!-- Generating test data that is big enough for STARsolo to detect and small enough | 465 <!-- Generating test data that is big enough for STARsolo to detect and small enough |
381 for Galaxy to test requires careful modification of input FASTA and GTF data, | 466 for Galaxy to test requires careful modification of input FASTA and GTF data, |
382 where the length of FASTA cannot exceed the largest position in the GTF file, | 467 where the length of FASTA cannot exceed the largest position in the GTF file, |
383 regardless of the FASTA starting sequence position. | 468 regardless of the FASTA starting sequence position. |
384 | 469 |
385 A full writeup of how to subset single cell data for use in STARsolo is given | 470 A full writeup of how to subset single cell data for use in STARsolo is given |
386 here: https://gist.github.com/mtekman/149a7c52fd73e5d8ebe49f5a27b0743d | 471 here: https://gist.github.com/mtekman/149a7c52fd73e5d8ebe49f5a27b0743d |
387 --> | 472 --> |
388 <tests> | 473 <tests> |
389 <test expect_num_outputs="6"> | 474 <test expect_num_outputs="7"> |
475 <!-- test 1 --> | |
390 <conditional name="refGenomeSource"> | 476 <conditional name="refGenomeSource"> |
391 <param name="geneSource" value="history" /> | 477 <param name="geneSource" value="history" /> |
392 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 478 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
393 <param name="genomeSAindexNbases" value="4" /> | 479 <param name="genomeSAindexNbases" value="4" /> |
394 <param name="sjdbOverhang" value="100" /> | 480 <param name="sjdbOverhang" value="100" /> |
401 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> | 487 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> |
402 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> | 488 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> |
403 </conditional> | 489 </conditional> |
404 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | 490 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> |
405 <conditional name="params"> | 491 <conditional name="params"> |
406 <param name="chemistry" value="CR3" /> | 492 <param name="chemistry" value="Cv3" /> |
407 </conditional> | 493 </conditional> |
408 <param name="soloUMIdedup" value="1MM_All" /> | 494 <conditional name="umidedup"> |
495 <param name="soloUMIdedup" value="1MM_All" /> | |
496 </conditional> | |
409 </conditional> | 497 </conditional> |
410 <section name="solo" > | 498 <section name="solo" > |
411 <conditional name="filter"> | 499 <conditional name="filter"> |
412 <param name="filter_type" value="no_filter" /> | 500 <param name="filter_type" value="no_filter" /> |
413 </conditional> | 501 </conditional> |
414 <param name="soloStrand" value="Forward" /> | 502 <param name="soloStrand" value="Forward" /> |
415 <param name="soloFeatures" value="Gene" /> | 503 <param name="soloFeatures" value="Gene" /> |
504 <param name="quantModeGene" value="true" /> | |
416 </section> | 505 </section> |
417 <output name="output_barcodes" > | 506 <output name="output_barcodes" > |
418 <assert_contents> | 507 <assert_contents> |
419 <!-- first and last line --> | 508 <!-- first and last line --> |
420 <has_line line="AAACCTGAGCGCTCCA" /> | 509 <has_line line="AAACCTGAGCGCTCCA" /> |
421 <has_line line="TTTGGTTAGTGGGCTA" /> | 510 <has_line line="TTTGGTTAGTGGGCTA" /> |
511 <has_n_lines n="394" /> | |
422 </assert_contents> | 512 </assert_contents> |
423 </output> | 513 </output> |
424 <output name="output_genes"> | 514 <output name="output_genes"> |
425 <assert_contents> | 515 <assert_contents> |
426 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> | 516 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> |
427 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> | 517 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> |
518 <has_n_lines n="14" /> | |
428 </assert_contents> | 519 </assert_contents> |
429 </output> | 520 </output> |
430 <output name="output_matrix" > | 521 <output name="output_matrix" > |
431 <assert_contents> | 522 <assert_contents> |
432 <has_line_matching expression="14\s+394\s+7" /> | 523 <has_line_matching expression="14\s+394\s+7" /> |
433 <has_line_matching expression="4\s+381\s+1" /> | 524 <has_line_matching expression="4\s+381\s+1" /> |
525 <has_n_lines n="10" /> | |
434 </assert_contents> | 526 </assert_contents> |
435 </output> | 527 </output> |
436 <output name="output_stats" > | 528 <output name="output_stats" > |
437 <assert_contents> | 529 <assert_contents> |
438 <has_line_matching expression="\s+nUnmapped\s+5823" /> | 530 <has_line_matching expression="\s+noUnmapped\s+5823" /> |
439 <has_line_matching expression="\s+nUMIs\s+8" /> | 531 <has_line_matching expression="\s+yesUMIs\s+8" /> |
440 </assert_contents> | 532 </assert_contents> |
441 </output> | 533 </output> |
442 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> | 534 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> |
535 <output name="reads_per_gene" > | |
536 <assert_contents> | |
537 <has_line_matching expression="ENSG00000279493\s+0\s+0\s+0" /> | |
538 <has_line_matching expression="ENSG00000275464\s+38\s+1\s+40" /> | |
539 </assert_contents> | |
540 </output> | |
443 </test> | 541 </test> |
444 <test expect_num_outputs="6"><!-- same as above, but using custom --> | 542 <test expect_num_outputs="6"> |
543 <!-- test 2 --> | |
544 <!-- same as above, but using custom and no reads_per_gene--> | |
445 <conditional name="refGenomeSource"> | 545 <conditional name="refGenomeSource"> |
446 <param name="geneSource" value="history" /> | 546 <param name="geneSource" value="history" /> |
447 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 547 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
448 <param name="genomeSAindexNbases" value="4" /> | 548 <param name="genomeSAindexNbases" value="4" /> |
449 <param name="sjdbOverhang" value="100" /> | 549 <param name="sjdbOverhang" value="100" /> |
469 <param name="soloStrand" value="Forward" /> | 569 <param name="soloStrand" value="Forward" /> |
470 <param name="soloFeatures" value="Gene" /> | 570 <param name="soloFeatures" value="Gene" /> |
471 </section> | 571 </section> |
472 <output name="output_barcodes_filtered" > | 572 <output name="output_barcodes_filtered" > |
473 <assert_contents> | 573 <assert_contents> |
574 <!-- first and last line --> | |
474 <has_line line="ACACCGGTCTAACGGT" /> | 575 <has_line line="ACACCGGTCTAACGGT" /> |
475 <has_line line="TTCTCAATCCACGTTC" /> | 576 <has_line line="TTCTCAATCCACGTTC" /> |
577 <has_n_lines n="7" /> | |
476 </assert_contents> | 578 </assert_contents> |
477 </output> | 579 </output> |
478 <output name="output_genes_filtered"> | 580 <output name="output_genes_filtered"> |
479 <assert_contents> | 581 <assert_contents> |
480 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> | 582 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> |
481 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> | 583 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> |
584 <has_n_lines n="14" /> | |
482 </assert_contents> | 585 </assert_contents> |
483 </output> | 586 </output> |
484 <output name="output_matrix_filtered" > | 587 <output name="output_matrix_filtered" > |
485 <assert_contents> | 588 <assert_contents> |
486 <has_line_matching expression="14\s+7\s+7" /> | 589 <has_line_matching expression="14\s+7\s+7" /> |
487 <has_line_matching expression="4\s+7\s+1" /> | 590 <has_line_matching expression="4\s+7\s+1" /> |
591 <has_n_lines n="10" /> | |
488 </assert_contents> | 592 </assert_contents> |
489 </output> | 593 </output> |
490 <output name="output_stats" > | 594 <output name="output_stats" > |
491 <assert_contents> | 595 <assert_contents> |
492 <has_line_matching expression="\s+nUnmapped\s+5823" /> | 596 <has_line_matching expression="\s+noUnmapped\s+5823" /> |
493 <has_line_matching expression="\s+nUMIs\s+8" /> | 597 <has_line_matching expression="\s+yesUMIs\s+8" /> |
494 </assert_contents> | 598 </assert_contents> |
495 </output> | 599 </output> |
496 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> | 600 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> |
497 </test> | 601 </test> |
498 <test expect_num_outputs="6"><!-- Multiple repeats test --> | 602 <test expect_num_outputs="6"> |
603 <!-- test 3 --> | |
604 <!-- Multiple repeats test --> | |
499 <conditional name="refGenomeSource"> | 605 <conditional name="refGenomeSource"> |
500 <param name="geneSource" value="history" /> | 606 <param name="geneSource" value="history" /> |
501 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 607 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
502 <param name="genomeSAindexNbases" value="4" /> | 608 <param name="genomeSAindexNbases" value="4" /> |
503 <param name="sjdbOverhang" value="100" /> | 609 <param name="sjdbOverhang" value="100" /> |
510 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz,pbmc_1k_v2_L001.R1.10k.fastq.gz,pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> | 616 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz,pbmc_1k_v2_L001.R1.10k.fastq.gz,pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> |
511 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz,pbmc_1k_v2_L001.R2.10k.fastq.gz,pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> | 617 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz,pbmc_1k_v2_L001.R2.10k.fastq.gz,pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> |
512 </conditional> | 618 </conditional> |
513 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | 619 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> |
514 <conditional name="params"> | 620 <conditional name="params"> |
515 <param name="chemistry" value="CR3" /> | 621 <param name="chemistry" value="Cv3" /> |
516 </conditional> | 622 </conditional> |
517 <param name="soloUMIdedup" value="1MM_All" /> | 623 <conditional name="umidedup"> |
624 <param name="soloUMIdedup" value="1MM_All" /> | |
625 </conditional> | |
518 </conditional> | 626 </conditional> |
519 <section name="solo" > | 627 <section name="solo" > |
520 <param name="soloStrand" value="Forward" /> | 628 <param name="soloStrand" value="Forward" /> |
521 <param name="soloFeatures" value="Gene" /> | 629 <param name="soloFeatures" value="Gene" /> |
522 </section> | 630 </section> |
523 <output name="output_barcodes_filtered" > | 631 <output name="output_barcodes_filtered" > |
524 <assert_contents> | 632 <assert_contents> |
525 <has_line line="ACACCGGTCTAACGGT" /> | 633 <has_line line="ACACCGGTCTAACGGT" /> |
526 <has_line line="TTCTCAATCCACGTTC" /> | 634 <has_line line="TTCTCAATCCACGTTC" /> |
527 </assert_contents> | 635 <has_n_lines n="7" /> |
528 </output> | 636 </assert_contents> |
529 <!-- BAM output is huge, we don't need to test here --> | 637 </output> |
638 <output name="output_BAM" > | |
639 <assert_contents> | |
640 <has_size value="166147" delta="600" /> | |
641 </assert_contents> | |
642 </output> | |
530 </test> | 643 </test> |
531 <test expect_num_outputs="6"> | 644 <test expect_num_outputs="10"> |
645 <!-- test 4 --> | |
532 <!-- Test with paired collection --> | 646 <!-- Test with paired collection --> |
533 <conditional name="refGenomeSource"> | 647 <conditional name="refGenomeSource"> |
534 <param name="geneSource" value="history" /> | 648 <param name="geneSource" value="history" /> |
535 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 649 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
536 <param name="genomeSAindexNbases" value="4" /> | 650 <param name="genomeSAindexNbases" value="4" /> |
548 </collection> | 662 </collection> |
549 </param> | 663 </param> |
550 </conditional> | 664 </conditional> |
551 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | 665 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> |
552 <conditional name="params"> | 666 <conditional name="params"> |
553 <param name="chemistry" value="CR3" /> | 667 <param name="chemistry" value="Cv3" /> |
554 </conditional> | 668 </conditional> |
555 <param name="soloUMIdedup" value="1MM_All" /> | 669 <conditional name="umidedup"> |
670 <param name="soloUMIdedup" value="1MM_All" /> | |
671 </conditional> | |
556 </conditional> | 672 </conditional> |
557 <section name="solo" > | 673 <section name="solo" > |
558 <param name="soloStrand" value="Forward" /> | 674 <param name="soloStrand" value="Forward" /> |
559 <param name="soloFeatures" value="Gene" /> | 675 <param name="soloFeatures" value="Gene" /> |
560 </section> | 676 </section> |
677 <conditional name="outWig"> | |
678 <param name="outWigType" value="bedGraph" /> | |
679 </conditional> | |
561 <output name="output_barcodes_filtered" > | 680 <output name="output_barcodes_filtered" > |
562 <assert_contents> | 681 <assert_contents> |
563 <has_line line="ACACCGGTCTAACGGT" /> | 682 <has_line line="ACACCGGTCTAACGGT" /> |
564 <has_line line="TTCTCAATCCACGTTC" /> | 683 <has_line line="TTCTCAATCCACGTTC" /> |
684 <has_n_lines n="7" /> | |
565 </assert_contents> | 685 </assert_contents> |
566 </output> | 686 </output> |
567 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> | 687 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> |
688 <output name="signal_unique_str1" file="Signal.Unique.str1.out.bg" /> | |
689 <output name="signal_uniquemultiple_str1" file="Signal.UniqueMultiple.str1.out.bg" /> | |
690 <output name="signal_unique_str2" file="Signal.Unique.str2.out.bg" /> | |
691 <output name="signal_uniquemultiple_str2" file="Signal.UniqueMultiple.str2.out.bg" /> | |
568 </test> | 692 </test> |
569 <test expect_num_outputs="6"> | 693 <test expect_num_outputs="9"> |
570 <!-- Test soloFeatures, soloCBmatchWLtype, soloCellFilter, soloOutFormatFeaturesGeneField3, soloUMIfiltering --> | 694 <!-- test 5 --> |
695 <!-- Test soloFeatures, soloCBmatchWLtype, soloCellFilter, soloOutFormatFeaturesGeneField3 --> | |
571 <conditional name="refGenomeSource"> | 696 <conditional name="refGenomeSource"> |
572 <param name="geneSource" value="history" /> | 697 <param name="geneSource" value="history" /> |
573 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 698 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
574 <param name="genomeSAindexNbases" value="4" /> | 699 <param name="genomeSAindexNbases" value="4" /> |
575 <param name="sjdbOverhang" value="100" /> | 700 <param name="sjdbOverhang" value="100" /> |
583 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> | 708 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> |
584 </conditional> | 709 </conditional> |
585 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | 710 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> |
586 <param name="soloCBmatchWLtype" value="1MM_multi_pseudocounts" /> | 711 <param name="soloCBmatchWLtype" value="1MM_multi_pseudocounts" /> |
587 <conditional name="params"> | 712 <conditional name="params"> |
588 <param name="chemistry" value="CR3" /> | 713 <param name="chemistry" value="Cv3" /> |
589 </conditional> | 714 </conditional> |
590 <param name="soloUMIdedup" value="1MM_All" /> | 715 <conditional name="umidedup"> |
716 <param name="soloUMIdedup" value="1MM_All" /> | |
717 </conditional> | |
591 </conditional> | 718 </conditional> |
592 <section name="solo" > | 719 <section name="solo" > |
593 <param name="soloUMIfiltering" value="MultiGeneUMI" /> | |
594 <param name="soloStrand" value="Forward" /> | 720 <param name="soloStrand" value="Forward" /> |
595 <param name="soloFeatures" value="GeneFull" /> | 721 <param name="soloFeatures" value="GeneFull" /> |
596 <conditional name="filter"> | 722 <conditional name="filter"> |
597 <param name="filter_type" value="topcells" /> | 723 <param name="filter_type" value="topcells" /> |
598 <param name="n_cells" value="5" /> | 724 <param name="n_cells" value="5" /> |
725 <param name="output_raw" value="true" /> | |
599 </conditional> | 726 </conditional> |
600 <param name="soloOutFormatFeaturesGeneField3" value="Dummy Text" /> | 727 <param name="soloOutFormatFeaturesGeneField3" value="Dummy Text" /> |
601 </section> | 728 </section> |
602 <output name="output_barcodes_filtered" > | 729 <output name="output_barcodes_filtered" > |
603 <assert_contents> | 730 <assert_contents> |
604 <!-- first and last line --> | 731 <!-- first and last line --> |
605 <has_line line="AGACGTTCAAGGCTCC" /> | 732 <has_line line="AGACGTTCAAGGCTCC" /> |
606 <has_line line="TCAACGAAGCTAGTGG" /> | 733 <has_line line="TCAACGAAGCTAGTGG" /> |
734 <has_n_lines n="6" /> | |
607 </assert_contents> | 735 </assert_contents> |
608 </output> | 736 </output> |
609 <output name="output_genes_filtered" > | 737 <output name="output_genes_filtered" > |
610 <assert_contents> | 738 <assert_contents> |
611 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Dummy\s+Text" /> | 739 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Dummy\s+Text" /> |
612 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Dummy\s+Text" /> | 740 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Dummy\s+Text" /> |
741 <has_n_lines n="14" /> | |
613 </assert_contents> | 742 </assert_contents> |
614 </output> | 743 </output> |
615 <output name="output_matrixGeneFull_filtered" > | 744 <output name="output_matrixGeneFull_filtered" > |
616 <assert_contents> | 745 <assert_contents> |
617 <has_line_matching expression="14\s+6\s+14" /> | 746 <has_line_matching expression="14\s+6\s+14" /> |
618 <has_line_matching expression="10\s+6\s+1" /> | 747 <has_line_matching expression="10\s+6\s+1" /> |
748 <has_n_lines n="17" /> | |
749 </assert_contents> | |
750 </output> | |
751 <output name="output_barcodes" > | |
752 <assert_contents> | |
753 <!-- first and last line --> | |
754 <has_line line="AAACCTGAGCGCTCCA" /> | |
755 <has_line line="TTTGGTTAGTGGGCTA" /> | |
756 <has_n_lines n="394" /> | |
757 </assert_contents> | |
758 </output> | |
759 <output name="output_genes"> | |
760 <assert_contents> | |
761 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Dummy\s+Text" /> | |
762 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Dummy\s+Text" /> | |
763 <has_n_lines n="14" /> | |
764 </assert_contents> | |
765 </output> | |
766 <output name="output_matrix" > | |
767 <assert_contents> | |
768 <has_line_matching expression="14\s+394\s+195" /> | |
769 <has_line_matching expression="3\s+1\s+1" /> | |
770 <has_n_lines n="198" /> | |
619 </assert_contents> | 771 </assert_contents> |
620 </output> | 772 </output> |
621 </test> | 773 </test> |
622 <test expect_num_outputs="6"> | 774 <test expect_num_outputs="6"> |
775 <!-- test 6 --> | |
623 <!-- Emptydrops filtering --> | 776 <!-- Emptydrops filtering --> |
624 <conditional name="refGenomeSource"> | 777 <conditional name="refGenomeSource"> |
625 <param name="geneSource" value="history" /> | 778 <param name="geneSource" value="history" /> |
626 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 779 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
627 <param name="genomeSAindexNbases" value="4" /> | 780 <param name="genomeSAindexNbases" value="4" /> |
635 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> | 788 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> |
636 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> | 789 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> |
637 </conditional> | 790 </conditional> |
638 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | 791 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> |
639 <conditional name="params"> | 792 <conditional name="params"> |
640 <param name="chemistry" value="CR3" /> | 793 <param name="chemistry" value="Cv3" /> |
641 </conditional> | 794 </conditional> |
642 <param name="soloUMIdedup" value="1MM_All" /> | 795 <conditional name="umidedup"> |
796 <param name="soloUMIdedup" value="1MM_All" /> | |
797 </conditional> | |
643 </conditional> | 798 </conditional> |
644 <section name="solo" > | 799 <section name="solo" > |
645 <conditional name="filter"> | 800 <conditional name="filter"> |
646 <param name="filter_type" value="emptydrops" /> | 801 <param name="filter_type" value="emptydrops" /> |
647 <param name="nExpectedCells" value="5" /> | 802 <param name="nExpectedCells" value="5" /> |
661 <output name="output_barcodes_filtered"> | 816 <output name="output_barcodes_filtered"> |
662 <assert_contents> | 817 <assert_contents> |
663 <!-- first and last line --> | 818 <!-- first and last line --> |
664 <has_line line="ACACCGGTCTAACGGT" /> | 819 <has_line line="ACACCGGTCTAACGGT" /> |
665 <has_line line="TTCTCAATCCACGTTC" /> | 820 <has_line line="TTCTCAATCCACGTTC" /> |
821 <has_n_lines n="7" /> | |
666 </assert_contents> | 822 </assert_contents> |
667 </output> | 823 </output> |
668 <output name="output_genes_filtered"> | 824 <output name="output_genes_filtered"> |
669 <assert_contents> | 825 <assert_contents> |
670 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> | 826 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> |
671 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> | 827 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> |
828 <has_n_lines n="14" /> | |
672 </assert_contents> | 829 </assert_contents> |
673 </output> | 830 </output> |
674 <output name="output_matrix_filtered" > | 831 <output name="output_matrix_filtered" > |
675 <assert_contents> | 832 <assert_contents> |
676 <has_line_matching expression="14\s+7\s+7" /> | 833 <has_line_matching expression="14\s+7\s+7" /> |
677 <has_line_matching expression="4\s+7\s+1" /> | 834 <has_line_matching expression="4\s+7\s+1" /> |
835 <has_n_lines n="10" /> | |
678 </assert_contents> | 836 </assert_contents> |
679 </output> | 837 </output> |
680 <output name="output_stats" > | 838 <output name="output_stats" > |
681 <assert_contents> | 839 <assert_contents> |
682 <has_line_matching expression="\s+nUnmapped\s+5823" /> | 840 <has_line_matching expression="\s+noUnmapped\s+5823" /> |
683 <has_line_matching expression="\s+nUMIs\s+8" /> | 841 <has_line_matching expression="\s+yesUMIs\s+8" /> |
684 </assert_contents> | 842 </assert_contents> |
685 </output> | 843 </output> |
686 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> | 844 <output name="output_BAM" value="filtered3.bam" compare="sim_size" delta="600" /> |
687 </test> | 845 </test> |
688 <test expect_num_outputs="6"> | 846 <test expect_num_outputs="6"> |
847 <!-- test 7 --> | |
689 <!-- Test soloType CB_UMI_Complex --> | 848 <!-- Test soloType CB_UMI_Complex --> |
690 <conditional name="refGenomeSource"> | 849 <conditional name="refGenomeSource"> |
691 <param name="geneSource" value="history" /> | 850 <param name="geneSource" value="history" /> |
692 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 851 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
693 <param name="genomeSAindexNbases" value="4" /> | 852 <param name="genomeSAindexNbases" value="4" /> |
720 <param name="umi_end_anchor" value="3" /> | 879 <param name="umi_end_anchor" value="3" /> |
721 <param name="umi_end_anchor_pos" value="14" /> | 880 <param name="umi_end_anchor_pos" value="14" /> |
722 <param name="soloAdapterSequence" value="GAGTGATTGCTTGTGACGCCTT" /> | 881 <param name="soloAdapterSequence" value="GAGTGATTGCTTGTGACGCCTT" /> |
723 <param name="soloAdapterMismatchesNmax" value="1" /> | 882 <param name="soloAdapterMismatchesNmax" value="1" /> |
724 <param name="clipAdapterType" value="CellRanger4" /> | 883 <param name="clipAdapterType" value="CellRanger4" /> |
725 <param name="soloUMIdedup" value="1MM_All" /> | 884 <conditional name="umidedup"> |
885 <param name="soloUMIdedup" value="1MM_All" /> | |
886 </conditional> | |
726 <param name="soloCBmatchWLtype" value="1MM" /> | 887 <param name="soloCBmatchWLtype" value="1MM" /> |
727 </conditional> | 888 </conditional> |
728 <output name="output_barcodes_filtered" > | 889 <output name="output_barcodes_filtered" > |
729 <assert_contents> | 890 <assert_contents> |
730 <!-- first and last line --> | 891 <!-- first and last line --> |
731 <has_line line="ACAACGTGG_AAACCTCC" /> | 892 <has_line line="ACAACGTGG_AAACCTCC" /> |
732 <has_line line="ATTCCAGAC_TTCGCTGG" /> | 893 <has_line line="ATTCCAGAC_TTCGCTGG" /> |
894 <has_n_lines n="33" /> | |
733 </assert_contents> | 895 </assert_contents> |
734 </output> | 896 </output> |
735 <output name="output_genes_filtered"> | 897 <output name="output_genes_filtered"> |
736 <assert_contents> | 898 <assert_contents> |
737 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> | 899 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> |
738 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> | 900 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> |
901 <has_n_lines n="14" /> | |
739 </assert_contents> | 902 </assert_contents> |
740 </output> | 903 </output> |
741 <output name="output_matrix_filtered" > | 904 <output name="output_matrix_filtered" > |
742 <assert_contents> | 905 <assert_contents> |
743 <has_line_matching expression="14\s+33\s+36" /> | 906 <has_line_matching expression="14\s+33\s+36" /> |
744 <has_line_matching expression="2\s+33\s+1" /> | 907 <has_line_matching expression="2\s+33\s+1" /> |
908 <has_n_lines n="39" /> | |
745 </assert_contents> | 909 </assert_contents> |
746 </output> | 910 </output> |
747 <output name="output_stats" > | 911 <output name="output_stats" > |
748 <assert_contents> | 912 <assert_contents> |
749 <has_line_matching expression="\s+nExactMatch\s+791" /> | 913 <has_line_matching expression="\s+yesWLmatchExact\s+791" /> |
750 <has_line_matching expression="\s+nUMIs\s+36" /> | 914 <has_line_matching expression="\s+yesUMIs\s+36" /> |
751 </assert_contents> | 915 </assert_contents> |
752 </output> | 916 </output> |
753 </test> | 917 </test> |
754 <test expect_num_outputs="6"> | 918 <test expect_num_outputs="6"> |
919 <!-- test 8 --> | |
755 <!-- Test soloType SmartSeq --> | 920 <!-- Test soloType SmartSeq --> |
756 <conditional name="refGenomeSource"> | 921 <conditional name="refGenomeSource"> |
757 <param name="geneSource" value="history" /> | 922 <param name="geneSource" value="history" /> |
758 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | 923 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> |
759 <param name="genomeSAindexNbases" value="4" /> | 924 <param name="genomeSAindexNbases" value="4" /> |
822 </element> | 987 </element> |
823 </collection> | 988 </collection> |
824 </param> | 989 </param> |
825 </conditional> | 990 </conditional> |
826 <param name="cell_ids" value="smartseq.cellids.txt" /> | 991 <param name="cell_ids" value="smartseq.cellids.txt" /> |
827 <param name="soloUMIdedup" value="Exact" /> | 992 <conditional name="umidedup"> |
993 <param name="soloUMIdedup" value="Exact" /> | |
994 </conditional> | |
828 </conditional> | 995 </conditional> |
829 <section name="solo" > | 996 <section name="solo" > |
830 <param name="soloStrand" value="Unstranded" /> | 997 <param name="soloStrand" value="Unstranded" /> |
831 <conditional name="filter"> | 998 <conditional name="filter"> |
832 <param name="filter_type" value="topcells" /> | 999 <param name="filter_type" value="topcells" /> |
835 </section> | 1002 </section> |
836 <output name="output_barcodes_filtered" > | 1003 <output name="output_barcodes_filtered" > |
837 <assert_contents> | 1004 <assert_contents> |
838 <has_line line="CSC6_D02" /> | 1005 <has_line line="CSC6_D02" /> |
839 <not_has_text text="MGH26_A02" /> | 1006 <not_has_text text="MGH26_A02" /> |
1007 <has_n_lines n="3" /> | |
840 </assert_contents> | 1008 </assert_contents> |
841 </output> | 1009 </output> |
842 <output name="output_genes_filtered"> | 1010 <output name="output_genes_filtered"> |
843 <assert_contents> | 1011 <assert_contents> |
844 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> | 1012 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> |
845 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> | 1013 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> |
1014 <has_n_lines n="14" /> | |
846 </assert_contents> | 1015 </assert_contents> |
847 </output> | 1016 </output> |
848 <output name="output_matrix_filtered" > | 1017 <output name="output_matrix_filtered" > |
849 <assert_contents> | 1018 <assert_contents> |
850 <has_line_matching expression="14\s+3\s+10" /> | 1019 <has_line_matching expression="14\s+3\s+10" /> |
851 <has_line_matching expression="12\s+3\s+1" /> | 1020 <has_line_matching expression="12\s+3\s+1" /> |
1021 <has_n_lines n="13" /> | |
852 </assert_contents> | 1022 </assert_contents> |
853 </output> | 1023 </output> |
854 <output name="output_stats" > | 1024 <output name="output_stats" > |
855 <assert_contents> | 1025 <assert_contents> |
856 <has_line_matching expression="\s+nExactMatch\s+9000" /> | 1026 <has_line_matching expression="\s+yesWLmatchExact\s+9000" /> |
857 <has_line_matching expression="\s+nUMIs\s+32" /> | 1027 <has_line_matching expression="\s+yesUMIs\s+32" /> |
1028 </assert_contents> | |
1029 </output> | |
1030 </test> | |
1031 <test expect_num_outputs="6"> | |
1032 <!-- test 9 --> | |
1033 <!-- Test outSAMattributes --> | |
1034 <conditional name="refGenomeSource"> | |
1035 <param name="geneSource" value="history" /> | |
1036 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | |
1037 <param name="genomeSAindexNbases" value="4" /> | |
1038 <param name="sjdbOverhang" value="100" /> | |
1039 <param name="sjdbGTFfile" value="filtered3.Homo_sapiens.GRCh38.100.chr21.gtf" ftype="gtf"/> | |
1040 </conditional> | |
1041 <conditional name="sc" > | |
1042 <param name="solo_type" value="CB_UMI_Simple" /> | |
1043 <conditional name="input_types"> | |
1044 <param name="use" value="repeat" /> | |
1045 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> | |
1046 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> | |
1047 </conditional> | |
1048 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | |
1049 <conditional name="params"> | |
1050 <param name="chemistry" value="Cv3" /> | |
1051 </conditional> | |
1052 <conditional name="umidedup"> | |
1053 <param name="soloUMIdedup" value="1MM_All" /> | |
1054 </conditional> | |
1055 </conditional> | |
1056 <section name="solo" > | |
1057 <conditional name="filter"> | |
1058 <param name="filter_type" value="no_filter" /> | |
1059 </conditional> | |
1060 <param name="soloStrand" value="Forward" /> | |
1061 <param name="soloFeatures" value="Gene" /> | |
1062 <param name="outSAMattributes" value="NH,HI,AS,nM,GX,GN,CB,UB" /> | |
1063 </section> | |
1064 <output name="output_barcodes" > | |
1065 <assert_contents> | |
1066 <!-- first and last line --> | |
1067 <has_line line="AAACCTGAGCGCTCCA" /> | |
1068 <has_line line="TTTGGTTAGTGGGCTA" /> | |
1069 <has_n_lines n="394" /> | |
1070 </assert_contents> | |
1071 </output> | |
1072 <output name="output_genes"> | |
1073 <assert_contents> | |
1074 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Gene\s+Expression" /> | |
1075 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Gene\s+Expression" /> | |
1076 <has_n_lines n="14" /> | |
1077 </assert_contents> | |
1078 </output> | |
1079 <output name="output_matrix" > | |
1080 <assert_contents> | |
1081 <has_line_matching expression="14\s+394\s+7" /> | |
1082 <has_line_matching expression="4\s+381\s+1" /> | |
1083 <has_n_lines n="10" /> | |
1084 </assert_contents> | |
1085 </output> | |
1086 <output name="output_stats" > | |
1087 <assert_contents> | |
1088 <has_line_matching expression="\s+noUnmapped\s+5823" /> | |
1089 <has_line_matching expression="\s+yesUMIs\s+8" /> | |
1090 </assert_contents> | |
1091 </output> | |
1092 <output name="output_BAM" > | |
1093 <assert_contents> | |
1094 <has_size value="153108" delta="600" /> | |
1095 </assert_contents> | |
1096 </output> | |
1097 </test> | |
1098 <test expect_num_outputs="6"> | |
1099 <!-- test 10 --> | |
1100 <!-- Test soloFeatures --> | |
1101 <conditional name="refGenomeSource"> | |
1102 <param name="geneSource" value="history" /> | |
1103 <param name="genomeFastaFiles" value="filtered3.Homo_sapiens.GRCh38.dna.chromosome.21.fa.gz" /> | |
1104 <param name="genomeSAindexNbases" value="4" /> | |
1105 <param name="sjdbOverhang" value="100" /> | |
1106 <param name="sjdbGTFfile" value="filtered3.Homo_sapiens.GRCh38.100.chr21.gtf" ftype="gtf"/> | |
1107 </conditional> | |
1108 <conditional name="sc" > | |
1109 <param name="solo_type" value="CB_UMI_Simple" /> | |
1110 <conditional name="input_types"> | |
1111 <param name="use" value="repeat" /> | |
1112 <param name="input1" value="pbmc_1k_v2_L001.R1.10k.fastq.gz" ftype="fastqsanger.gz" /> | |
1113 <param name="input2" value="pbmc_1k_v2_L001.R2.10k.fastq.gz" ftype="fastqsanger.gz" /> | |
1114 </conditional> | |
1115 <param name="soloCBwhitelist" value="filtered.barcodes.txt" /> | |
1116 <param name="soloCBmatchWLtype" value="1MM_multi_pseudocounts" /> | |
1117 <conditional name="params"> | |
1118 <param name="chemistry" value="Cv3" /> | |
1119 </conditional> | |
1120 <conditional name="umidedup"> | |
1121 <param name="soloUMIdedup" value="1MM_CR" /> | |
1122 <param name="soloUMIfiltering" value="MultiGeneUMI" /> | |
1123 </conditional> | |
1124 </conditional> | |
1125 <section name="solo" > | |
1126 <param name="soloStrand" value="Forward" /> | |
1127 <param name="soloFeatures" value="GeneFull_ExonOverIntron" /> | |
1128 <conditional name="filter"> | |
1129 <param name="filter_type" value="no_filter" /> | |
1130 </conditional> | |
1131 <param name="soloOutFormatFeaturesGeneField3" value="Dummy Text" /> | |
1132 </section> | |
1133 <output name="output_barcodes" > | |
1134 <assert_contents> | |
1135 <!-- first and last line --> | |
1136 <has_line line="AAACCTGAGCGCTCCA" /> | |
1137 <has_line line="TTTGGTTAGTGGGCTA" /> | |
1138 <has_n_lines n="394" /> | |
1139 </assert_contents> | |
1140 </output> | |
1141 <output name="output_genes" > | |
1142 <assert_contents> | |
1143 <has_line_matching expression="ENSG00000279493\s+FP565260\.4\s+Dummy\s+Text" /> | |
1144 <has_line_matching expression="ENSG00000279064\s+FP236315\.1\s+Dummy\s+Text" /> | |
1145 <has_n_lines n="14" /> | |
1146 </assert_contents> | |
1147 </output> | |
1148 <output name="output_matrixGeneFull" > | |
1149 <assert_contents> | |
1150 <has_line_matching expression="14\s+394\s+104" /> | |
1151 <has_line_matching expression="10\s+2\s+1" /> | |
1152 <has_n_lines n="107" /> | |
858 </assert_contents> | 1153 </assert_contents> |
859 </output> | 1154 </output> |
860 </test> | 1155 </test> |
861 </tests> | 1156 </tests> |
862 <help><![CDATA[ | 1157 <help><![CDATA[ |