Mercurial > repos > iuc > snapatac2_preprocessing

comparison preprocessing.xml @ 1:cec3e76eaf05 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/snapatac2 commit 1e34deee1e39c0c65e1e29a9d28becc7aaf23a4f
author iuc
date Thu, 23 May 2024 15:20:02 +0000
parents 00a6721e1f81
children 48d9421bf176
comparison
equal deleted inserted replaced
0:00a6721e1f81 1:cec3e76eaf05
20 @CMD_read_inputs@ 20 @CMD_read_inputs@
21 #end if 21 #end if
22 22
23 #if $method.method == 'pp.make_fragment_file' 23 #if $method.method == 'pp.make_fragment_file'
24 sa.pp.make_fragment_file( 24 sa.pp.make_fragment_file(
25 bam_file = '$method.bam_file', 25 bam_file = '$method.bam_file',
26 is_paired = $method.is_paired, 26 is_paired = $method.is_paired,
27 #if $method.barcode.extract_type == 'from_tag' 27 #if $method.barcode.extract_type == 'from_tag'
28 #if $method.barcode.barcode_tag != '' 28 #if $method.barcode.barcode_tag != ''
29 barcode_tag = '$method.barcode.barcode_tag', 29 barcode_tag = '$method.barcode.barcode_tag',
30 #end if 30 #end if
31 #elif $method.barcode.extract_type == 'from_read_names' 31 #elif $method.barcode.extract_type == 'from_read_names'
32 #if $method.barcode.barcode_regex != '' 32 #if $method.barcode.barcode_regex != ''
33 barcode_regex = '$method.barcode.barcode_regex', 33 barcode_regex = '$method.barcode.barcode_regex',
34 #end if 34 #end if
35 #end if 35 #end if
36 #if $method.umi_tag != '' 36 #if $method.umi_tag != ''
37 umi_tag = '$method.umi_tag', 37 umi_tag = '$method.umi_tag',
38 #end if 38 #end if
39 #if $method.umi_regex != '' 39 #if $method.umi_regex != ''
40 umi_regex = '$method.umi_regex', 40 umi_regex = '$method.umi_regex',
41 #end if 41 #end if
42 shift_right = $method.shift_right, 42 shift_right = $method.shift_right,
43 shift_left = $method.shift_left, 43 shift_left = $method.shift_left,
44 min_mapq = $method.min_mapq, 44 min_mapq = $method.min_mapq,
45 chunk_size = $method.chunk_size, 45 chunk_size = $method.chunk_size,
46 compression = 'gzip', 46 compression = 'gzip',
47 output_file = '$fragments_out', 47 output_file = '$fragments_out',
48 tempdir = "." 48 tempdir = "."
49 ) 49 )
50 50
51 #else if $method.method == 'pp.import_data' 51 #else if $method.method == 'pp.import_data'
52 import csv 52 import csv
53 with open('$method.chrom_sizes') as f: 53 with open('$method.chrom_sizes') as f:
54 chr_sizes = {x[0]:int(x[1]) for x in csv.reader(f, delimiter='\t')} 54 chr_sizes = {x[0]:int(x[1]) for x in csv.reader(f, delimiter='\t')}
55 55
56 sa.pp.import_data( 56 sa.pp.import_data(
57 fragment_file = '$method.fragment_file', 57 fragment_file = '$method.fragment_file',
58 chrom_sizes = chr_sizes, 58 chrom_sizes = chr_sizes,
59 min_num_fragments = $method.min_num_fragments, 59 min_num_fragments = $method.min_num_fragments,
60 sorted_by_barcode = $method.sorted_by_barcode, 60 sorted_by_barcode = $method.sorted_by_barcode,
61 #if str($method.whitelist) != 'None' 61 #if str($method.whitelist) != 'None'
62 whitelist = '$method.whitelist', 62 whitelist = '$method.whitelist',
63 #end if 63 #end if
64 shift_left = $method.shift_left, 64 shift_left = $method.shift_left,
65 shift_right = $method.shift_right, 65 shift_right = $method.shift_right,
66 #set $chr_mt = ([x.strip() for x in str($method.chrM).split(',')]) 66 #set $chr_mt = ([x.strip() for x in str($method.chrM).split(',')])
67 chrM = $chr_mt, 67 chrM = $chr_mt,
68 chunk_size = $method.chunk_size, 68 chunk_size = $method.chunk_size,
69 file = 'anndata.h5ad', 69 file = 'anndata.h5ad',
70 n_jobs = os.getenv("GALAXY_SLOTS", 4) 70 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
71 ) 71 )
72 72
73 #else if $method.method == 'pp.add_tile_matrix' 73 #else if $method.method == 'pp.add_tile_matrix'
74 sa.pp.add_tile_matrix( 74 sa.pp.add_tile_matrix(
75 adata, 75 adata,
76 bin_size = $method.bin_size, 76 bin_size = $method.bin_size,
77 inplace = True, 77 chunk_size = $method.chunk_size,
78 chunk_size = $method.chunk_size, 78 #if $method.exclude_chroms != ''
79 #if $method.exclude_chroms != '' 79 #set $excl_chroms = ([x.strip() for x in str($method.exclude_chroms).split(',')])
80 #set $excl_chroms = ([x.strip() for x in str($method.exclude_chroms).split(',')]) 80 exclude_chroms = $excl_chroms,
81 exclude_chroms = $excl_chroms, 81 #end if
82 #end if 82 #if $method.min_frag_size
83 #if $method.min_frag_size 83 min_frag_size = $method.min_frag_size,
84 min_frag_size = $method.min_frag_size, 84 #end if
85 #end if 85 #if $method.max_frag_size
86 #if $method.max_frag_size 86 max_frag_size = $method.max_frag_size,
87 max_frag_size = $method.max_frag_size, 87 #end if
88 #end if 88 ##counting_strategy = '$method.counting_strategy',
89 ##counting_strategy = '$method.counting_strategy', 89 count_frag_as_reads = $method.count_frag_as_reads,
90 count_frag_as_reads = $method.count_frag_as_reads, 90 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
91 n_jobs = os.getenv("GALAXY_SLOTS", 4)
92 ) 91 )
93 92
94 #else if $method.method == 'pp.make_gene_matrix' 93 #else if $method.method == 'pp.make_gene_matrix'
95 sa.pp.make_gene_matrix( 94 sa.pp.make_gene_matrix(
96 adata, 95 adata,
97 gene_anno = '$method.gene_anno', 96 gene_anno = '$method.gene_anno',
98 inplace = True, 97 chunk_size = $method.chunk_size,
99 chunk_size = $method.chunk_size, 98 use_x = $method.use_x,
100 use_x = $method.use_x, 99 id_type = '$method.id_type',
101 id_type = '$method.id_type', 100 transcript_name_key = '$method.transcript_name_key',
102 transcript_name_key = '$method.transcript_name_key', 101 transcript_id_key = '$method.transcript_id_key',
103 transcript_id_key = '$method.transcript_id_key', 102 gene_name_key = '$method.gene_name_key',
104 gene_name_key = '$method.gene_name_key', 103 gene_id_key = '$method.gene_id_key',
105 gene_id_key = '$method.gene_id_key', 104 #if $method.min_frag_size
106 #if $method.min_frag_size 105 min_frag_size = $method.min_frag_size,
107 min_frag_size = $method.min_frag_size, 106 #end if
108 #end if 107 #if $method.max_frag_size
109 #if $method.max_frag_size 108 max_frag_size = $method.max_frag_size,
110 max_frag_size = $method.max_frag_size, 109 #end if
111 #end if 110 ##counting_strategy = '$method.counting_strategy'
112 ##counting_strategy = '$method.counting_strategy' 111 count_frag_as_reads = $method.count_frag_as_reads
113 count_frag_as_reads = $method.count_frag_as_reads
114 ) 112 )
115 113
116 #else if $method.method == 'pp.filter_cells' 114 #else if $method.method == 'pp.filter_cells'
117 sa.pp.filter_cells( 115 sa.pp.filter_cells(
118 adata, 116 adata,
119 min_counts = $method.min_counts, 117 min_counts = $method.min_counts,
120 min_tsse = $method.min_tsse, 118 min_tsse = $method.min_tsse,
121 #if $method.max_counts 119 #if $method.max_counts
122 max_counts = $method.max_counts, 120 max_counts = $method.max_counts,
123 #end if 121 #end if
124 #if $method.max_tsse 122 #if $method.max_tsse
125 max_tsse = $method.max_tsse, 123 max_tsse = $method.max_tsse,
126 #end if 124 #end if
127 inplace = True, 125 inplace = True,
128 n_jobs = os.getenv("GALAXY_SLOTS", 4) 126 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
129 ) 127 )
130 128
131 #else if $method.method == 'pp.select_features' 129 #else if $method.method == 'pp.select_features'
132 sa.pp.select_features( 130 sa.pp.select_features(
133 adata, 131 adata,
134 n_features = $method.n_features, 132 n_features = $method.n_features,
135 filter_lower_quantile = $method.filter_lower_quantile, 133 filter_lower_quantile = $method.filter_lower_quantile,
136 filter_upper_quantile = $method.filter_upper_quantile, 134 filter_upper_quantile = $method.filter_upper_quantile,
137 #if str($method.whitelist) != 'None' 135 #if str($method.whitelist) != 'None'
138 whitelist = '$method.whitelist', 136 whitelist = '$method.whitelist',
139 #end if 137 #end if
140 #if str($method.blacklist) != 'None' 138 #if str($method.blacklist) != 'None'
141 blacklist = '$method.blacklist', 139 blacklist = '$method.blacklist',
142 #end if 140 #end if
143 max_iter = $method.max_iter, 141 max_iter = $method.max_iter,
144 inplace = True, 142 inplace = True,
145 n_jobs = os.getenv("GALAXY_SLOTS", 4) 143 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
146 ) 144 )
147 145
148 #else if $method.method == 'pp.scrublet' 146 #else if $method.method == 'pp.scrublet'
149 sa.pp.scrublet( 147 sa.pp.scrublet(
150 adata, 148 adata,
151 #if $method.features 149 #if $method.features
152 features = '$method.features', 150 features = '$method.features',
153 #end if 151 #end if
154 n_comps = $method.n_comps, 152 n_comps = $method.n_comps,
155 sim_doublet_ratio = $method.sim_doublet_ratio, 153 sim_doublet_ratio = $method.sim_doublet_ratio,
156 expected_doublet_rate = $method.expected_doublet_rate, 154 expected_doublet_rate = $method.expected_doublet_rate,
157 #if $method.n_neighbors 155 #if $method.n_neighbors
158 n_neighbors = $method.n_neighbors, 156 n_neighbors = $method.n_neighbors,
159 #end if 157 #end if
160 use_approx_neighbors = $method.use_approx_neighbors, 158 use_approx_neighbors = $method.use_approx_neighbors,
161 random_state = $method.random_state, 159 random_state = $method.random_state,
162 inplace = True, 160 inplace = True,
163 n_jobs = os.getenv("GALAXY_SLOTS", 4) 161 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
164 ) 162 )
165 163
166 #else if $method.method == 'pp.filter_doublets' 164 #else if $method.method == 'pp.filter_doublets'
167 sa.pp.filter_doublets( 165 sa.pp.filter_doublets(
168 adata, 166 adata,
169 #if $method.probability_threshold 167 #if $method.probability_threshold
170 probability_threshold = $method.probability_threshold, 168 probability_threshold = $method.probability_threshold,
171 #end if 169 #end if
172 #if $method.score_threshold 170 #if $method.score_threshold
173 score_threshold = $method.score_threshold, 171 score_threshold = $method.score_threshold,
174 #end if 172 #end if
175 inplace = True, 173 inplace = True,
176 n_jobs = os.getenv("GALAXY_SLOTS", 4) 174 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
177 ) 175 )
178 176
179 #else if $method.method == 'pp.mnc_correct' 177 #else if $method.method == 'pp.mnc_correct'
180 sa.pp.mnc_correct( 178 sa.pp.mnc_correct(
181 adata, 179 adata,
182 batch = '$method.batch', 180 batch = '$method.batch',
183 n_neighbors = $method.n_neighbors, 181 n_neighbors = $method.n_neighbors,
184 n_clusters = $method.n_clusters, 182 n_clusters = $method.n_clusters,
185 n_iter = $method.n_iter, 183 n_iter = $method.n_iter,
186 @CMD_params_data_integration@ 184 @CMD_params_data_integration@
187 inplace = True, 185 inplace = True,
188 n_jobs = os.getenv("GALAXY_SLOTS", 4) 186 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
189 ) 187 )
190 188
191 #else if $method.method == 'pp.harmony' 189 #else if $method.method == 'pp.harmony'
192 sa.pp.harmony( 190 sa.pp.harmony(
193 adata, 191 adata,
194 batch = '$method.batch', 192 batch = '$method.batch',
195 @CMD_params_data_integration@ 193 @CMD_params_data_integration@
196 inplace = True 194 inplace = True
197 ) 195 )
198 196
199 #else if $method.method == 'pp.scanorama_integrate' 197 #else if $method.method == 'pp.scanorama_integrate'
200 sa.pp.scanorama_integrate( 198 sa.pp.scanorama_integrate(
201 adata, 199 adata,
202 batch = '$method.batch', 200 batch = '$method.batch',
203 n_neighbors = $method.n_neighbors, 201 n_neighbors = $method.n_neighbors,
204 @CMD_params_data_integration@ 202 @CMD_params_data_integration@
205 inplace = True 203 inplace = True
206 ) 204 )
207 205
208 #else if $method.method == 'metrics.frag_size_distr' 206 #else if $method.method == 'metrics.frag_size_distr'
209 sa.metrics.frag_size_distr( 207 sa.metrics.frag_size_distr(
210 adata, 208 adata,
211 max_recorded_size = $method.max_recorded_size, 209 max_recorded_size = $method.max_recorded_size,
212 add_key = '$method.add_key', 210 add_key = '$method.add_key',
213 inplace = True, 211 inplace = True,
214 n_jobs = os.getenv("GALAXY_SLOTS", 4) 212 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
215 ) 213 )
216 214
217 #else if $method.method == 'metrics.tsse' 215 #else if $method.method == 'metrics.tsse'
218 sa.metrics.tsse( 216 sa.metrics.tsse(
219 adata, 217 adata,
220 gene_anno = '$method.gene_anno', 218 gene_anno = '$method.gene_anno',
221 inplace = True, 219 inplace = True,
222 n_jobs = os.getenv("GALAXY_SLOTS", 4) 220 n_jobs = int(os.getenv("GALAXY_SLOTS", 4))
223 ) 221 )
224 #end if 222 #end if
225 223
226 #if $method.method != 'pp.make_fragment_file' and $method.method != 'pp.import_data' 224 #if $method.method != 'pp.make_fragment_file' and $method.method != 'pp.import_data'
227 @CMD_anndata_write_outputs@ 225 @CMD_anndata_write_outputs@
228 #end if 226 #end if
229 ]]></configfile> 227 ]]></configfile>
230 </configfiles> 228 </configfiles>
231 <inputs> 229 <inputs>
232 <conditional name="method"> 230 <conditional name="method">
233 <param name="method" type="select" label="Method used for preprocessing"> 231 <param name="method" type="select" label="Method used for preprocessing">
234 <option value="pp.make_fragment_file">Convert a BAM file to a fragment file, using 'pp.make_fragment_file'</option> 232 <option value="pp.make_fragment_file">Convert a BAM file to a fragment file, using 'pp.make_fragment_file'</option>
255 </param> 253 </param>
256 <when value="from_tag"> 254 <when value="from_tag">
257 <param argument="barcode_tag" type="text" value="CB" optional="true" label="Extract barcodes from TAG fields of BAM records"/> 255 <param argument="barcode_tag" type="text" value="CB" optional="true" label="Extract barcodes from TAG fields of BAM records"/>
258 </when> 256 </when>
259 <when value="from_read_names"> 257 <when value="from_read_names">
260 <param argument="barcode_regex" type="text" value="" optional="true" label="Extract barcodes from read names of BAM records using regular expressions" help="`(..:..:..:..):w+$` extracts `bd:69:Y6:10` from `A01535:24:HW2MMDSX2:2:1359:8513:3458:bd:69:Y6:10:TGATAGGTT``"/> 258 <param argument="barcode_regex" type="text" value="" optional="true" label="Extract barcodes from read names of BAM records using regular expressions" help="`(..:..:..:..):\w+$` extracts `bd:69:Y6:10` from `A01535:24:HW2MMDSX2:2:1359:8513:3458:bd:69:Y6:10:TGATAGGTT``"/>
261 </when> 259 </when>
262 </conditional> 260 </conditional>
263 <param argument="umi_tag" type="text" value="" optional="true" label="Extract UMI from TAG fields of BAM records"/> 261 <param argument="umi_tag" type="text" value="" optional="true" label="Extract UMI from TAG fields of BAM records"/>
264 <param argument="umi_regex" type="text" value="" optional="true" label="Extract UMI from read names of BAM records using regular expressions"/> 262 <param argument="umi_regex" type="text" value="" optional="true" label="Extract UMI from read names of BAM records using regular expressions"/>
265 <expand macro="param_shift"/> 263 <expand macro="param_shift"/>
391 <tests> 389 <tests>
392 <test expect_num_outputs="1"> 390 <test expect_num_outputs="1">
393 <!-- pp.make_fragment_file --> 391 <!-- pp.make_fragment_file -->
394 <conditional name="method"> 392 <conditional name="method">
395 <param name="method" value="pp.make_fragment_file"/> 393 <param name="method" value="pp.make_fragment_file"/>
396 <param name="bam_file" location="https://zenodo.org/records/11199963/files/pbmc_500_chr21_subsample.bam"/> 394 <param name="bam_file" location="https://zenodo.org/records/11260316/files/pbmc_500_chr21_subsample.bam"/>
397 <param name="is_paired" value="true"/> 395 <param name="is_paired" value="true"/>
398 <conditional name="barcode"> 396 <conditional name="barcode">
399 <param name="extract_type" value="from_tag"/> 397 <param name="extract_type" value="from_tag"/>
400 <param name="barcode_tag" value="CB"/> 398 <param name="barcode_tag" value="CB"/>
401 </conditional> 399 </conditional>
402 <param name="shift_left" value="4"/> 400 <param name="shift_left" value="4"/>
403 <param name="shift_right" value="-5"/> 401 <param name="shift_right" value="-5"/>
404 <param name="min_mapq" value="10"/> 402 <param name="min_mapq" value="10"/>
405 <param name="chunk_size" value="50000000"/> 403 <param name="chunk_size" value="50000000"/>
406 </conditional> 404 </conditional>
407 <output name="fragments_out" location="https://zenodo.org/records/11199963/files/pp.make_fragment_file.pbmc_500_chr21.tsv.gz" ftype="interval" compare="sim_size" delta_frac="0.1"/> 405 <output name="fragments_out" location="https://zenodo.org/records/11260316/files/pp.make_fragment_file.pbmc_500_chr21.tsv.gz" ftype="interval" compare="sim_size" delta_frac="0.1"/>
408 </test> 406 </test>
409 <test expect_num_outputs="2"> 407 <test expect_num_outputs="2">
410 <!-- pp.pp.import_data --> 408 <!-- pp.pp.import_data -->
411 <conditional name="method"> 409 <conditional name="method">
412 <param name="method" value="pp.import_data"/> 410 <param name="method" value="pp.import_data"/>
413 <param name="fragment_file" location="https://zenodo.org/records/11199963/files/pbmc_500_chr21.tsv.gz"/> 411 <param name="fragment_file" location="https://zenodo.org/records/11260316/files/pbmc_500_chr21.tsv.gz"/>
414 <param name="chrom_sizes" location="https://zenodo.org/records/11199963/files/chr21_size.tabular"/> 412 <param name="chrom_sizes" location="https://zenodo.org/records/11260316/files/chr21_size.tabular"/>
415 <param name="min_num_fragments" value="1"/> 413 <param name="min_num_fragments" value="1"/>
416 <param name="sorted_by_barcode" value="False"/> 414 <param name="sorted_by_barcode" value="False"/>
417 <param name="shift_left" value="0"/> 415 <param name="shift_left" value="0"/>
418 <param name="chrM" value="chrM, M"/> 416 <param name="chrM" value="chrM, M"/>
419 <param name="shift_right" value="0"/> 417 <param name="shift_right" value="0"/>
431 <has_text_matching expression="chrM = \['chrM', 'M'\]"/> 429 <has_text_matching expression="chrM = \['chrM', 'M'\]"/>
432 <has_text_matching expression="shift_right = 0"/> 430 <has_text_matching expression="shift_right = 0"/>
433 <has_text_matching expression="chunk_size = 1000"/> 431 <has_text_matching expression="chunk_size = 1000"/>
434 </assert_contents> 432 </assert_contents>
435 </output> 433 </output>
436 <output name="anndata_out" location="https://zenodo.org/records/11199963/files/pp.import_data.pbmc_500_chr21.h5ad" ftype="h5ad" compare="sim_size" delta_frac="0.1"/> 434 <output name="anndata_out" location="https://zenodo.org/records/11260316/files/pp.import_data.pbmc_500_chr21.h5ad" ftype="h5ad" compare="sim_size" delta_frac="0.1"/>
435 </test>
436 <test expect_num_outputs="2">
437 <!-- pp.make_gene_matrix -->
438 <conditional name="method">
439 <param name="method" value="pp.make_gene_matrix"/>
440 <param name="adata" location="https://zenodo.org/records/11260316/files/tl.leiden.modularity.pbmc_500_chr21.h5ad"/>
441 <param name="gene_anno" location="https://zenodo.org/records/11260316/files/chr21.gff3.gz"/>
442 <param name="chunk_size" value="500"/>
443 <param name="use_x" value="False"/>
444 <param name="id_type" value="gene"/>
445 <param name="transcript_name_key" value="transcript_name"/>
446 <param name="transcript_id_key" value="transcript_id"/>
447 <param name="gene_name_key" value="gene_name"/>
448 <param name="gene_id_key" value="gene_id"/>
449 <param name="count_frag_as_reads" value="True"/>
450 </conditional>
451 <section name="advanced_common">
452 <param name="show_log" value="true" />
453 </section>
454 <output name="hidden_output">
455 <assert_contents>
456 <has_text_matching expression="sa.pp.make_gene_matrix"/>
457 <has_text_matching expression="chunk_size = 500"/>
458 <has_text_matching expression="use_x = False"/>
459 <has_text_matching expression="id_type = 'gene'"/>
460 <has_text_matching expression="transcript_name_key = 'transcript_name'"/>
461 <has_text_matching expression="transcript_id_key = 'transcript_id'"/>
462 <has_text_matching expression="gene_name_key = 'gene_name'"/>
463 <has_text_matching expression="gene_id_key = 'gene_id'"/>
464 <has_text_matching expression="count_frag_as_reads = True"/>
465 </assert_contents>
466 </output>
467 <output name="anndata_out" location="https://zenodo.org/records/11260316/files/pp.make_gene_matrix.pbmc_500_chr21.h5ad" ftype="h5ad" compare="sim_size" delta_frac="0.1" />
468 </test>
469 <test expect_num_outputs="2">
470 <!-- metrics.tsse -->
471 <conditional name="method">
472 <param name="method" value="metrics.tsse"/>
473 <param name="adata" location="https://zenodo.org/records/11260316/files/pp.import_data.pbmc_500_chr21.h5ad"/>
474 <param name="gene_anno" location="https://zenodo.org/records/11260316/files/chr21.gff3.gz"/>
475 </conditional>
476 <section name="advanced_common">
477 <param name="show_log" value="true" />
478 </section>
479 <output name="hidden_output">
480 <assert_contents>
481 <has_text_matching expression="sa.metrics.tsse"/>
482 </assert_contents>
483 </output>
484 <output name="anndata_out" location="https://zenodo.org/records/11260316/files/metrics.tsse.pbmc_500_chr21.h5ad" ftype="h5ad" compare="sim_size" delta_frac="0.1" />
485 </test>
486 <test expect_num_outputs="2">
487 <!-- pp.filter_cells -->
488 <conditional name="method">
489 <param name="method" value="pp.filter_cells"/>
490 <param name="adata" location="https://zenodo.org/records/11260316/files/metrics.tsse.pbmc_500_chr21.h5ad"/>
491 <param name="min_counts" value="200"/>
492 <param name="min_tsse" value="5"/>
493 <param name="max_counts" value="10000"/>
494 </conditional>
495 <section name="advanced_common">
496 <param name="show_log" value="true" />
497 </section>
498 <output name="hidden_output">
499 <assert_contents>
500 <has_text_matching expression="sa.pp.filter_cells"/>
501 <has_text_matching expression="min_counts = 200"/>
502 <has_text_matching expression="min_tsse = 5"/>
503 <has_text_matching expression="max_counts = 10000"/>
504 </assert_contents>
505 </output>
506 <output name="anndata_out" location="https://zenodo.org/records/11260316/files/pp.filter_cells.pbmc_500_chr21.h5ad" ftype="h5ad" compare="sim_size" delta_frac="0.1" />
507 </test>
508 <test expect_num_outputs="2">
509 <!-- pp.add_tile_matrix -->
510 <conditional name="method">
511 <param name="method" value="pp.add_tile_matrix"/>
512 <param name="adata" location="https://zenodo.org/records/11260316/files/pp.filter_cells.pbmc_500_chr21.h5ad"/>
513 <param name="bin_size" value="5000"/>
514 <param name="chunk_size" value="500"/>
515 <param name="exclude_chroms" value="chr1, chr2, chr3, chr4, chr5, chr6, chr7, chr8, chr9, chr10, chr11, chr12, chr13, chr14, chr15, chr16, chr17, chr18, chr19, chr20, chr22, chrX, chrY"/>
516 <param name="count_frag_as_reads" value="True"/>
517 </conditional>
518 <section name="advanced_common">
519 <param name="show_log" value="true" />
520 </section>
521 <output name="hidden_output">
522 <assert_contents>
523 <has_text_matching expression="sa.pp.add_tile_matrix"/>
524 <has_text_matching expression="bin_size = 5000"/>
525 <has_text_matching expression="chunk_size = 500"/>
526 <has_text_matching expression="exclude_chroms = \['chr1', 'chr2', 'chr3', 'chr4', 'chr5', 'chr6', 'chr7', 'chr8', 'chr9', 'chr10', 'chr11', 'chr12', 'chr13', 'chr14', 'chr15', 'chr16', 'chr17', 'chr18', 'chr19', 'chr20', 'chr22', 'chrX', 'chrY'\]"/>
527 <has_text_matching expression="count_frag_as_reads = True"/>
528 </assert_contents>
529 </output>
530 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.add_tile_matrix.pbmc_500_chr21.h5ad"/>
531 </test>
532 <test expect_num_outputs="2">
533 <!-- pp.select_features -->
534 <conditional name="method">
535 <param name="method" value="pp.select_features"/>
536 <param name="adata" location="https://zenodo.org/records/11260316/files/pp.add_tile_matrix.pbmc_500_chr21.h5ad"/>
537 <param name="n_features" value="15000"/>
538 </conditional>
539 <section name="advanced_common">
540 <param name="show_log" value="true" />
541 </section>
542 <output name="hidden_output">
543 <assert_contents>
544 <has_text_matching expression="sa.pp.select_features"/>
545 <has_text_matching expression="n_features = 15000"/>
546 </assert_contents>
547 </output>
548 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.select_features.pbmc_500_chr21.h5ad"/>
549 </test>
550 <test expect_num_outputs="2">
551 <!-- pp.scrublet -->
552 <conditional name="method">
553 <param name="method" value="pp.scrublet"/>
554 <param name="adata" location="https://zenodo.org/records/11260316/files/pp.select_features.pbmc_500_chr21.h5ad"/>
555 <param name="n_comps" value="15"/>
556 <param name="sim_doublet_ratio" value="2.0"/>
557 <param name="expected_doublet_rate" value="0.1"/>
558 <param name="random_state" value="0"/>
559 </conditional>
560 <section name="advanced_common">
561 <param name="show_log" value="true" />
562 </section>
563 <output name="hidden_output">
564 <assert_contents>
565 <has_text_matching expression="sa.pp.scrublet"/>
566 <has_text_matching expression="n_comps = 15"/>
567 <has_text_matching expression="sim_doublet_ratio = 2.0"/>
568 <has_text_matching expression="expected_doublet_rate = 0.1"/>
569 <has_text_matching expression="random_state = 0"/>
570 </assert_contents>
571 </output>
572 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.scrublet.pbmc_500_chr21.h5ad"/>
573 </test>
574 <test expect_num_outputs="2">
575 <!-- pp.filter_doublets -->
576 <conditional name="method">
577 <param name="method" value="pp.filter_doublets"/>
578 <param name="adata" location="https://zenodo.org/records/11260316/files/pp.scrublet.pbmc_500_chr21.h5ad"/>
579 <param name="probability_threshold" value="0.1"/>
580 </conditional>
581 <section name="advanced_common">
582 <param name="show_log" value="true" />
583 </section>
584 <output name="hidden_output">
585 <assert_contents>
586 <has_text_matching expression="sa.pp.filter_doublets"/>
587 </assert_contents>
588 </output>
589 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.filter_doublets.pbmc_500_chr21.h5ad"/>
590 </test>
591 <test expect_num_outputs="2">
592 <!-- pp.mnc_correct -->
593 <conditional name="method">
594 <param name="method" value="pp.mnc_correct"/>
595 <param name="adata" location="https://zenodo.org/records/11260316/files/pbmc_500_chr21.batch.h5ad"/>
596 <param name="batch" value="batch"/>
597 <param name="n_neighbors" value="3"/>
598 <param name="n_clusters" value="10"/>
599 <param name="use_rep" value="X_spectral"/>
600 </conditional>
601 <section name="advanced_common">
602 <param name="show_log" value="true" />
603 </section>
604 <output name="hidden_output">
605 <assert_contents>
606 <has_text_matching expression="sa.pp.mnc_correct"/>
607 <has_text_matching expression="batch = 'batch'"/>
608 <has_text_matching expression="n_neighbors = 3"/>
609 <has_text_matching expression="n_clusters = 10"/>
610 <has_text_matching expression="batch = 'batch'"/>
611 <has_text_matching expression="use_rep = 'X_spectral'"/>
612 </assert_contents>
613 </output>
614 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.mnc_correct.pbmc_500_chr21.h5ad"/>
615 </test>
616 <test expect_num_outputs="2">
617 <!-- pp.harmony -->
618 <conditional name="method">
619 <param name="method" value="pp.harmony"/>
620 <param name="adata" location="https://zenodo.org/records/11260316/files/pbmc_500_chr21.batch.h5ad"/>
621 <param name="batch" value="batch"/>
622 <param name="use_rep" value="X_spectral"/>
623 </conditional>
624 <section name="advanced_common">
625 <param name="show_log" value="true" />
626 </section>
627 <output name="hidden_output">
628 <assert_contents>
629 <has_text_matching expression="sa.pp.harmony"/>
630 <has_text_matching expression="batch = 'batch'"/>
631 <has_text_matching expression="use_rep = 'X_spectral'"/>
632 </assert_contents>
633 </output>
634 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.harmony.pbmc_500_chr21.h5ad"/>
635 </test>
636 <test expect_num_outputs="2">
637 <!-- pp.scanorama_integrate -->
638 <conditional name="method">
639 <param name="method" value="pp.scanorama_integrate"/>
640 <param name="adata" location="https://zenodo.org/records/11260316/files/pbmc_500_chr21.batch.h5ad"/>
641 <param name="batch" value="batch"/>
642 <param name="use_rep" value="X_spectral"/>
643 </conditional>
644 <section name="advanced_common">
645 <param name="show_log" value="true" />
646 </section>
647 <output name="hidden_output">
648 <assert_contents>
649 <has_text_matching expression="sa.pp.scanorama_integrate"/>
650 <has_text_matching expression="batch = 'batch'"/>
651 <has_text_matching expression="use_rep = 'X_spectral'"/>
652 </assert_contents>
653 </output>
654 <output name="anndata_out" ftype="h5ad" compare="sim_size" delta_frac="0.1" location="https://zenodo.org/records/11260316/files/pp.scanorama_integrate.pbmc_500_chr21.h5ad"/>
655 </test>
656 <test expect_num_outputs="2">
657 <!-- metrics.frag_size_distr -->
658 <conditional name="method">
659 <param name="method" value="metrics.frag_size_distr"/>
660 <param name="adata" location="https://zenodo.org/records/11260316/files/pp.import_data.pbmc_500_chr21.h5ad"/>
661 <param name="max_recorded_size" value="500"/>
662 <param name="add_key" value="frag_size_distr"/>
663 </conditional>
664 <section name="advanced_common">
665 <param name="show_log" value="true" />
666 </section>
667 <output name="hidden_output">
668 <assert_contents>
669 <has_text_matching expression="sa.metrics.frag_size_distr"/>
670 <has_text_matching expression="add_key = 'frag_size_distr'"/>
671 </assert_contents>
672 </output>
673 <output name="anndata_out" location="https://zenodo.org/records/11260316/files/metrics.frag_size_distr.pbmc_500_chr21.h5ad" ftype="h5ad" compare="sim_size" delta_frac="0.1" />
437 </test> 674 </test>
438 </tests> 675 </tests>
439 <help><![CDATA[ 676 <help><![CDATA[
440 Convert a BAM file`to a fragment file, `using pp.make_fragment_file` 677 Convert a BAM file`to a fragment file, using `pp.make_fragment_file`
441 ==================================================================== 678 ====================================================================
442 679
443 Convert a BAM file to a fragment file. 680 Convert a BAM file to a fragment file.
444 681
445 Convert a BAM file to a fragment file by performing the following steps: 682 Convert a BAM file to a fragment file by performing the following steps: