Mercurial > repos > iuc > meryl
comparison meryl.xml @ 0:068920e730f4 draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/meryl commit f94b7a1d482e29e10c33205ae5394c4c4a1e74e0"
| author | iuc |
|---|---|
| date | Mon, 05 Apr 2021 15:04:36 +0000 |
| parents | |
| children | e5b18909f73d |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:068920e730f4 |
|---|---|
| 1 <tool id='meryl' name='Meryl' version='@TOOL_VERSION@+@GALAXY_TOOL_VERSION@' profile='20.01'> | |
| 2 <description>a genomic k-mer counter and sequence utility</description> | |
| 3 <macros> | |
| 4 <import>macros.xml</import> | |
| 5 </macros> | |
| 6 <expand macro='edam_ontology' /> | |
| 7 <expand macro='requirements' /> | |
| 8 <version_command>meryl --version</version_command> | |
| 9 <command detect_errors='exit_code'><![CDATA[ | |
| 10 #if $operation_type.command_type == 'count-kmers' | |
| 11 meryl | |
| 12 #if $operation_type.options_kmer_size.kmer_size == 'estimate' | |
| 13 #from math import log | |
| 14 $operation_type.options_kmer_size.input_kmer_size = int(log(int($operation_type.options_kmer_size.genome_size)*(1-float($operation_type.options_kmer_size.collision_rate))/float($operation_type.options_kmer_size.collision_rate))/log(4)) | |
| 15 #end if | |
| 16 $operation_type.count_operations | |
| 17 k=$operation_type.options_kmer_size.input_kmer_size | |
| 18 $operation_type.input_reads | |
| 19 output read-db.meryl && | |
| 20 echo 'K-mer size: ${$operation_type.options_kmer_size.input_kmer_size}' && | |
| 21 tar -zcf read-db.meryldb read-db.meryl | |
| 22 #elif $operation_type.command_type == 'filter-kmers' | |
| 23 mkdir -p ./temp_db/ && | |
| 24 tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ && | |
| 25 mv ./temp_db/* tmp.meryl && | |
| 26 meryl | |
| 27 $operation_type.filter_operations | |
| 28 $operation_type.N | |
| 29 tmp.meryl | |
| 30 output read-db.meryl && | |
| 31 tar -zcf read-db.meryldb read-db.meryl | |
| 32 #elif $operation_type.command_type == 'arithmetic-kmers' | |
| 33 mkdir -p ./temp_db/ && | |
| 34 tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ && | |
| 35 mv ./temp_db/* tmp.meryl && | |
| 36 meryl | |
| 37 $operation_type.arithmetic_operations | |
| 38 $operation_type.X | |
| 39 tmp.meryl | |
| 40 output read-db.meryl && | |
| 41 tar -zcf read-db.meryldb read-db.meryl | |
| 42 #elif $operation_type.command_type == 'groups-kmers' | |
| 43 mkdir -p ./temp_db/ && | |
| 44 tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ && | |
| 45 mv ./temp_db/* tmp_01.meryl && | |
| 46 tar -zxf $operation_type.input_meryldb_03 -C ./temp_db/ && | |
| 47 mv ./temp_db/* tmp_02.meryl && | |
| 48 meryl | |
| 49 $operation_type.groups_operations | |
| 50 tmp_01.meryl | |
| 51 tmp_02.meryl | |
| 52 output read-db.meryl && | |
| 53 tar -zcf read-db.meryldb read-db.meryl | |
| 54 #elif $operation_type.command_type == 'histogram-kmers' | |
| 55 mkdir -p ./temp_db/ && | |
| 56 tar -zxf $operation_type.input_meryldb_02 -C ./temp_db/ && | |
| 57 mv ./temp_db/* tmp.meryl && | |
| 58 meryl histogram tmp.meryl > read-db.hist | |
| 59 #elif $operation_type.command_type == 'trio-mode' | |
| 60 export MERQURY=\$(dirname \$(command -v merqury.sh))/../share/merqury/ && | |
| 61 #for $i, $read in enumerate($paternal_reads): | |
| 62 mkdir 'paternal{$i}.meryl' && | |
| 63 meryl count k=$operation_type.options_kmer_size.input_kmer_size '${read}' output 'paternal{$i}.meryl' && | |
| 64 #end for | |
| 65 meryl union-sum paternal*.meryl output pat.meryl && | |
| 66 #for $i, $read in enumerate($maternal_reads): | |
| 67 mkdir 'maternal{$i}.meryl' && | |
| 68 meryl count k=$operation_type.options_kmer_size.input_kmer_size '${read}' output 'maternal{$i}.meryl' && | |
| 69 #end for | |
| 70 meryl union-sum maternal*.meryl output mat.meryl && | |
| 71 #for $i, $read in enumerate($child_reads): | |
| 72 mkdir 'child{$i}.meryl' && | |
| 73 meryl count k=$operation_type.options_kmer_size.input_kmer_size '${read}' output 'child{$i}.meryl' && | |
| 74 #end for | |
| 75 meryl union-sum child*.meryl output child.meryl && | |
| 76 ## mat specific kmers | |
| 77 meryl difference mat.meryl pat.meryl output mat.only.meryl && | |
| 78 meryl histogram mat.only.meryl > mat.only.hist && | |
| 79 java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar mat.only.hist > mat.only.ploidy && | |
| 80 VAR=`sed -n 2p mat.only.ploidy | awk '{print \$NF}'` && | |
| 81 meryl greater-than \$VAR output mat.only.filt.meryl mat.only.meryl && | |
| 82 | |
| 83 ## pat specific kmers | |
| 84 meryl difference pat.meryl mat.meryl output pat.only.meryl && | |
| 85 meryl histogram pat.only.meryl > pat.only.hist && | |
| 86 java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar pat.only.hist > pat.only.ploidy && | |
| 87 VAR=`sed -n 2p pat.only.ploidy | awk '{print \$NF}'` && | |
| 88 meryl greater-than \$VAR output pat.only.filt.meryl pat.only.meryl && | |
| 89 | |
| 90 ## shared kmers | |
| 91 meryl intersect output shared.meryl mat.meryl pat.meryl && | |
| 92 | |
| 93 ## mat hapmers | |
| 94 meryl intersect output mat.inherited.meryl child.meryl mat.only.filt.meryl && | |
| 95 meryl histogram mat.inherited.meryl > mat.inherited.hist && | |
| 96 java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar mat.inherited.hist > mat.inherited.ploidy && | |
| 97 VAR=`sed -n 2p mat.inherited.ploidy | awk '{print \$NF}'` && | |
| 98 meryl greater-than \$VAR output mat.hapmer.meryl mat.inherited.meryl && | |
| 99 tar -czf 'mat.meryldb' mat.hapmer.meryl && | |
| 100 | |
| 101 ## pat hapmers | |
| 102 meryl intersect output pat.inherited.meryl child.meryl pat.only.filt.meryl && | |
| 103 meryl histogram pat.inherited.meryl > pat.inherited.hist && | |
| 104 java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar pat.inherited.hist > pat.inherited.ploidy && | |
| 105 VAR=`sed -n 2p pat.inherited.ploidy | awk '{print \$NF}'` && | |
| 106 meryl greater-than \$VAR output pat.hapmer.meryl pat.inherited.meryl && | |
| 107 tar -czf 'pat.meryldb' pat.hapmer.meryl && | |
| 108 | |
| 109 ## shared hapmers | |
| 110 meryl intersect output shared.inherited.meryl child.meryl shared.meryl && | |
| 111 meryl histogram shared.inherited.meryl > shared.inherited.hist && | |
| 112 java -jar -Xmx1g \$MERQURY/eval/kmerHistToPloidyDepth.jar shared.inherited.hist > shared.inherited.ploidy && | |
| 113 VAR=`sed -n 2p shared.inherited.ploidy | awk '{print \$NF}'` && | |
| 114 meryl greater-than \$VAR output shared.filt.meryl shared.inherited.meryl && | |
| 115 | |
| 116 ## child hapmers | |
| 117 meryl union-sum output child.inherited.meryl mat.inherited.meryl pat.inherited.meryl shared.inherited.meryl && | |
| 118 meryl difference output read.only.meryl child.meryl child.inherited.meryl && | |
| 119 tar -czf 'read-db.meryldb' read.only.meryl && | |
| 120 echo 'K-mer size: ${$operation_type.options_kmer_size.input_kmer_size}' | |
| 121 #end if | |
| 122 ]]> </command> | |
| 123 <inputs> | |
| 124 <conditional name="operation_type"> | |
| 125 <param name="command_type" type="select" label="Operation type selector" help="Select a type of operation"> | |
| 126 <option value="count-kmers">Count operations</option> | |
| 127 <option value="filter-kmers">Filter operations</option> | |
| 128 <option value="arithmetic-kmers">Arithmetic operations on kmer counts</option> | |
| 129 <option value="groups-kmers">Operations on sets of k-mers</option> | |
| 130 <option value="histogram-kmers">Generate histogram dataset</option> | |
| 131 <option value="trio-mode">Build hap-mer dbs for trios</option> | |
| 132 </param> | |
| 133 <when value="count-kmers"> | |
| 134 <param name="count_operations" type="select" label="Count operations" help="Select an operation to be executed"> | |
| 135 <option value="count">Count: count the occurrences of canonical k-mers</option> | |
| 136 <option value="count-forward">Count-forward: count the occurreces of forward k-mers</option> | |
| 137 <option value="count-reverse">Count-reverse: count the occurreces of reverse k-mers</option> | |
| 138 </param> | |
| 139 <param name="input_reads" type="data" format="fastq,fasta" label="Input sequences" help="Select your reads in FASTA/FASTQ format." /> | |
| 140 <conditional name="options_kmer_size"> | |
| 141 <param name="kmer_size" type="select" label="K-mer size selector"> | |
| 142 <option value="provide">Set a k-mer size</option> | |
| 143 <option value="estimate">Estimate the best k-mer size</option> | |
| 144 </param> | |
| 145 <when value="provide"> | |
| 146 <param name="input_kmer_size" type="integer" min="1" max="50" value="" label="K-mer size" help="Haploid genome size or diploid genome size, depending on what we evaluate. In bp. Only required if the k-mer size is not provided." /> | |
| 147 </when> | |
| 148 <when value="estimate"> | |
| 149 <param name="genome_size" type="integer" min="1000" max="70000000000" value="1000" label="Genome size" help="Haploid genome size or diploid genome size, depending on what we evaluate. In bp. Only required if the k-mer size is not provided." /> | |
| 150 <param name="collision_rate" type="float" min="0.0001" max="0.01" value="0.001" label="Tolerable collision rate" help="Tolerable collision rate. By default is 0.001." /> | |
| 151 </when> | |
| 152 </conditional> | |
| 153 </when> | |
| 154 <when value="filter-kmers"> | |
| 155 <param name="filter_operations" type="select" label="Filter operations" help="Select an operation to be executed"> | |
| 156 <option value="less-than">Less-than: return k-mers that occur fewer than N times in the input</option> | |
| 157 <option value="greater-than">Greater-than: return k-mers that occur more than N times in the input</option> | |
| 158 <option value="equal-to">Equal-to: return k-mers that occur exactly N times in the input</option> | |
| 159 <option value="not-equal-to">Not-equal-to: return k-mers that do not occur exactly N times in the input</option> | |
| 160 </param> | |
| 161 <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" help="Select a meryldb dataset" /> | |
| 162 <param name="N" type="integer" min="0" max="50000000000" value="" optional="true" label="Filter value" /> | |
| 163 </when> | |
| 164 <when value="arithmetic-kmers"> | |
| 165 <param name="arithmetic_operations" type="select" label="Arithmetic operations" help="Select an operation to be executed"> | |
| 166 <option value="increase">Increase: add x to the count of each k-mer</option> | |
| 167 <option value="decrease">Decrease: subsctract x from the count of each k-mer</option> | |
| 168 <option value="multiply">Multiply: multiply the count of each k-mer by x</option> | |
| 169 <option value="divide">Divide: divide the count of each k-mer by x</option> | |
| 170 <option value="divide-round">Divide-round: divide the count of each k-mer by x and round th results</option> | |
| 171 <option value="modulo">Modulo: set the count of each k-mer to the remainder of the count divided by x</option> | |
| 172 </param> | |
| 173 <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" help="Select a meryldb dataset" /> | |
| 174 <param name="X" type="integer" min="1" max="1000000" value="" optional="true" label="Operand" /> | |
| 175 </when> | |
| 176 <when value="groups-kmers"> | |
| 177 <param name="groups_operations" type="select" label="Operations on sets of kmers" help="Select an operation to be executed"> | |
| 178 <option value="union">Union: return k-mers that occur in any input</option> | |
| 179 <option value="union-min">Union-min: return k-mers that occur in any input, set the count to the minimum count</option> | |
| 180 <option value="union-max">Union-max: return k-mers that occur in any input, set the count to the maximum count</option> | |
| 181 <option value="union-sum">Union-sum: return k-mers that occur in any input, set the count to the sum of the counts</option> | |
| 182 <option value="intersect">Intersect: return k-mers that occur in all inputs, set the count to the count in the first input</option> | |
| 183 <option value="intersect-min">Intersect-min: return k-mers that occur in all inputs, set the count to the minimum count</option> | |
| 184 <option value="intersect-max">Intersect-max: return k-mers that occur in all inputs, set the count to the maximum count</option> | |
| 185 <option value="intersect-sum">Intersect-sum: return k-mers that occur in all inputs, set the count to the sum of the counts</option> | |
| 186 <option value="subtract">Subtract: return k-mers that occur in the first input, subtracting counts from the other inputs</option> | |
| 187 <option value="difference">Difference: return k-mers that occur in the first input, but none of the other inputs</option> | |
| 188 <option value="symmetric-difference">Symmetric-difference: return k-mers that occur in exactly one input</option> | |
| 189 </param> | |
| 190 <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" /> | |
| 191 <param name="input_meryldb_03" type="data" format="meryldb" label="Input meryldb" /> | |
| 192 </when> | |
| 193 <when value="histogram-kmers"> | |
| 194 <param name="input_meryldb_02" type="data" format="meryldb" label="Input meryldb" /> | |
| 195 </when> | |
| 196 <when value="trio-mode"> | |
| 197 <param name="child_reads" type="data" format="fastq,fasta" multiple="true" label="F1 reads" help="Select F1 reads used for generating the assembly" /> | |
| 198 <param name="paternal_reads" type="data" format="fastq,fasta" multiple="true" optional="true" label="Paternal reads" help="Select the paternal reads used for generating the assembly" /> | |
| 199 <param name="maternal_reads" type="data" format="fastq,fasta" multiple="true" optional="true" label="Maternal reads" help="Select the maternal reads used for generating the assembly" /> | |
| 200 <conditional name="options_kmer_size"> | |
| 201 <param name="kmer_size" type="select" label="K-mer size selector"> | |
| 202 <option value="provide">Set a k-mer size</option> | |
| 203 <option value="estimate">Estimate the best k-mer size</option> | |
| 204 </param> | |
| 205 <when value="provide"> | |
| 206 <param name="input_kmer_size" type="integer" min="1" max="50" value="" label="K-mer size" help="Haploid genome size or diploid genome size, depending on what we evaluate. In bp. Only required if the k-mer size is not provided." /> | |
| 207 </when> | |
| 208 <when value="estimate"> | |
| 209 <param name="genome_size" type="integer" min="1000" max="70000000000" value="1000" label="Genome size" help="Haploid genome size or diploid genome size, depending on what we evaluate. In bp. Only required if the k-mer size is not provided." /> | |
| 210 <param name="collision_rate" type="float" min="0.0001" max="0.01" value="0.001" label="Tolerable collision rate" help="Tolerable collision rate. By default is 0.001." /> | |
| 211 </when> | |
| 212 </conditional> | |
| 213 </when> | |
| 214 </conditional> | |
| 215 </inputs> | |
| 216 <outputs> | |
| 217 <data name="read_db" format="meryldb" from_work_dir="read-db.meryldb" label="${tool.name} on ${on_string}: read-db.meryldb"> | |
| 218 <filter>operation_type["command_type"] != "histogram-kmers"</filter> | |
| 219 </data> | |
| 220 <data name="read_db_hist" format="tabular" from_work_dir="read-db.hist" label="${tool.name} on ${on_string}: read-db histogram"> | |
| 221 <filter>operation_type["command_type"] == "histogram-kmers" or operation_type["command_type"] == "trio-mode"</filter> | |
| 222 </data> | |
| 223 <data name="pat_db" format="meryldb" from_work_dir="pat.meryldb" label="${tool.name} on ${on_string}: read-db.meryl"> | |
| 224 <filter>operation_type["command_type"] == "trio-mode"</filter> | |
| 225 </data> | |
| 226 <data name="pat_db_hist" format="tabular" from_work_dir="pat.inherited.hist" label="${tool.name} on ${on_string}: paternal inherited histogram"> | |
| 227 <filter>operation_type["command_type"] == "trio-mode"</filter> | |
| 228 </data> | |
| 229 <data name="mat_db" format="meryldb" from_work_dir="mat.meryldb" label="${tool.name} on ${on_string}: mat.meryl"> | |
| 230 <filter>operation_type["command_type"] == "trio-mode"</filter> | |
| 231 </data> | |
| 232 <data name="mat_db_hist" format="tabular" from_work_dir="mat.inherited.hist" label="${tool.name} on ${on_string}: maternal inherited histogram"> | |
| 233 <filter>operation_type["command_type"] == "trio-mode"</filter> | |
| 234 </data> | |
| 235 </outputs> | |
| 236 <tests> | |
| 237 <test expect_num_outputs="1"> | |
| 238 <conditional name="operation_type"> | |
| 239 <param name="command_type" value="count-kmers" /> | |
| 240 <param name="count_operation" value="count" /> | |
| 241 <param name="input_reads" value="child.fasta" /> | |
| 242 <conditional name="options_kmer_size"> | |
| 243 <param name="kmer_size" value="provide" /> | |
| 244 <param name="input_kmer_size" value="7" /> | |
| 245 </conditional> | |
| 246 </conditional> | |
| 247 <output name="read_db" ftype="meryldb"> | |
| 248 <assert_contents> | |
| 249 <has_size value="22152" delta="300" /> | |
| 250 </assert_contents> | |
| 251 </output> | |
| 252 <assert_stdout> | |
| 253 <has_line line="K-mer size: 7" /> | |
| 254 </assert_stdout> | |
| 255 </test> | |
| 256 <test expect_num_outputs="1"> | |
| 257 <conditional name="operation_type"> | |
| 258 <param name="command_type" value="filter-kmers" /> | |
| 259 <param name="filter_operations" value="less-than" /> | |
| 260 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 261 <param name="N" value="100" /> | |
| 262 </conditional> | |
| 263 <output name="read_db" ftype="meryldb"> | |
| 264 <assert_contents> | |
| 265 <has_size value="32077" delta="300" /> | |
| 266 </assert_contents> | |
| 267 </output> | |
| 268 </test> | |
| 269 <test expect_num_outputs="1"> | |
| 270 <conditional name="operation_type"> | |
| 271 <param name="command_type" value="filter-kmers" /> | |
| 272 <param name="filter_operations" value="greater-than" /> | |
| 273 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 274 <param name="N" value="80" /> | |
| 275 </conditional> | |
| 276 <output name="read_db" ftype="meryldb"> | |
| 277 <assert_contents> | |
| 278 <has_size value="49951" delta="300" /> | |
| 279 </assert_contents> | |
| 280 </output> | |
| 281 </test> | |
| 282 <test expect_num_outputs="1"> | |
| 283 <conditional name="operation_type"> | |
| 284 <param name="command_type" value="filter-kmers" /> | |
| 285 <param name="filter_operations" value="equal-to" /> | |
| 286 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 287 <param name="N" value="100" /> | |
| 288 </conditional> | |
| 289 <output name="read_db" ftype="meryldb"> | |
| 290 <assert_contents> | |
| 291 <has_size value="2621" delta="300" /> | |
| 292 </assert_contents> | |
| 293 </output> | |
| 294 </test> | |
| 295 <test expect_num_outputs="1"> | |
| 296 <conditional name="operation_type"> | |
| 297 <param name="command_type" value="filter-kmers" /> | |
| 298 <param name="filter_operations" value="not-equal-to" /> | |
| 299 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 300 <param name="N" value="100" /> | |
| 301 </conditional> | |
| 302 <output name="read_db" ftype="meryldb"> | |
| 303 <assert_contents> | |
| 304 <has_size value="59378" delta="300" /> | |
| 305 </assert_contents> | |
| 306 </output> | |
| 307 </test> | |
| 308 <test expect_num_outputs="1"> | |
| 309 <conditional name="operation_type"> | |
| 310 <param name="command_type" value="arithmetic-kmers" /> | |
| 311 <param name="arithmetic_operations" value="increase" /> | |
| 312 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 313 <param name="X" value="100000" /> | |
| 314 </conditional> | |
| 315 <output name="read_db" ftype="meryldb"> | |
| 316 <assert_contents> | |
| 317 <has_size value="59822" delta="300" /> | |
| 318 </assert_contents> | |
| 319 </output> | |
| 320 </test> | |
| 321 <test expect_num_outputs="1"> | |
| 322 <conditional name="operation_type"> | |
| 323 <param name="command_type" value="arithmetic-kmers" /> | |
| 324 <param name="arithmetic_operations" value="decrease" /> | |
| 325 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 326 <param name="X" value="100" /> | |
| 327 </conditional> | |
| 328 <output name="read_db" ftype="meryldb"> | |
| 329 <assert_contents> | |
| 330 <has_size value="42625" delta="300" /> | |
| 331 </assert_contents> | |
| 332 </output> | |
| 333 </test> | |
| 334 <test expect_num_outputs="1"> | |
| 335 <conditional name="operation_type"> | |
| 336 <param name="command_type" value="arithmetic-kmers" /> | |
| 337 <param name="arithmetic_operations" value="multiply" /> | |
| 338 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 339 <param name="X" value="3" /> | |
| 340 </conditional> | |
| 341 <output name="read_db" ftype="meryldb"> | |
| 342 <assert_contents> | |
| 343 <has_size value="60832" delta="300" /> | |
| 344 </assert_contents> | |
| 345 </output> | |
| 346 </test> | |
| 347 <test expect_num_outputs="1"> | |
| 348 <conditional name="operation_type"> | |
| 349 <param name="command_type" value="arithmetic-kmers" /> | |
| 350 <param name="arithmetic_operations" value="divide" /> | |
| 351 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 352 <param name="X" value="2" /> | |
| 353 </conditional> | |
| 354 <output name="read_db" ftype="meryldb"> | |
| 355 <assert_contents> | |
| 356 <has_size value="56569" delta="300" /> | |
| 357 </assert_contents> | |
| 358 </output> | |
| 359 </test> | |
| 360 <test expect_num_outputs="1"> | |
| 361 <conditional name="operation_type"> | |
| 362 <param name="command_type" value="arithmetic-kmers" /> | |
| 363 <param name="arithmetic_operations" value="divide-round" /> | |
| 364 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 365 <param name="X" value="2" /> | |
| 366 </conditional> | |
| 367 <output name="read_db" ftype="meryldb"> | |
| 368 <assert_contents> | |
| 369 <has_size value="56539" delta="300" /> | |
| 370 </assert_contents> | |
| 371 </output> | |
| 372 </test> | |
| 373 <!-- test 11 --> | |
| 374 <test expect_num_outputs="1"> | |
| 375 <conditional name="operation_type"> | |
| 376 <param name="command_type" value="arithmetic-kmers" /> | |
| 377 <param name="arithmetic_operations" value="modulo" /> | |
| 378 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 379 <param name="X" value="3" /> | |
| 380 </conditional> | |
| 381 <output name="read_db" ftype="meryldb"> | |
| 382 <assert_contents> | |
| 383 <has_size value="37542" delta="300" /> | |
| 384 </assert_contents> | |
| 385 </output> | |
| 386 </test> | |
| 387 <test expect_num_outputs="1"> | |
| 388 <conditional name="operation_type"> | |
| 389 <param name="command_type" value="groups-kmers" /> | |
| 390 <param name="groups_operations" value="union" /> | |
| 391 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 392 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 393 </conditional> | |
| 394 <output name="read_db" ftype="meryldb"> | |
| 395 <assert_contents> | |
| 396 <has_size value="36417" delta="300" /> | |
| 397 </assert_contents> | |
| 398 </output> | |
| 399 </test> | |
| 400 <test expect_num_outputs="1"> | |
| 401 <conditional name="operation_type"> | |
| 402 <param name="command_type" value="groups-kmers" /> | |
| 403 <param name="groups_operations" value="union-min" /> | |
| 404 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 405 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 406 </conditional> | |
| 407 <output name="read_db" ftype="meryldb"> | |
| 408 <assert_contents> | |
| 409 <has_size value="59226" delta="300" /> | |
| 410 </assert_contents> | |
| 411 </output> | |
| 412 </test> | |
| 413 <test expect_num_outputs="1"> | |
| 414 <conditional name="operation_type"> | |
| 415 <param name="command_type" value="groups-kmers" /> | |
| 416 <param name="groups_operations" value="union-max" /> | |
| 417 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 418 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 419 </conditional> | |
| 420 <output name="read_db" ftype="meryldb"> | |
| 421 <assert_contents> | |
| 422 <has_size value="59234" delta="300" /> | |
| 423 </assert_contents> | |
| 424 </output> | |
| 425 </test> | |
| 426 <test expect_num_outputs="1"> | |
| 427 <conditional name="operation_type"> | |
| 428 <param name="command_type" value="groups-kmers" /> | |
| 429 <param name="groups_operations" value="union-sum" /> | |
| 430 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 431 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 432 </conditional> | |
| 433 <output name="read_db" ftype="meryldb"> | |
| 434 <assert_contents> | |
| 435 <has_size value="59004" delta="300" /> | |
| 436 </assert_contents> | |
| 437 </output> | |
| 438 </test> | |
| 439 <test expect_num_outputs="1"> | |
| 440 <conditional name="operation_type"> | |
| 441 <param name="command_type" value="groups-kmers" /> | |
| 442 <param name="groups_operations" value="intersect" /> | |
| 443 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 444 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 445 </conditional> | |
| 446 <output name="read_db" ftype="meryldb"> | |
| 447 <assert_contents> | |
| 448 <has_size value="14951" delta="300" /> | |
| 449 </assert_contents> | |
| 450 </output> | |
| 451 </test> | |
| 452 <test expect_num_outputs="1"> | |
| 453 <conditional name="operation_type"> | |
| 454 <param name="command_type" value="groups-kmers" /> | |
| 455 <param name="groups_operations" value="intersect-min" /> | |
| 456 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 457 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 458 </conditional> | |
| 459 <output name="read_db" ftype="meryldb"> | |
| 460 <assert_contents> | |
| 461 <has_size value="14957" delta="300" /> | |
| 462 </assert_contents> | |
| 463 </output> | |
| 464 </test> | |
| 465 <test expect_num_outputs="1"> | |
| 466 <conditional name="operation_type"> | |
| 467 <param name="command_type" value="groups-kmers" /> | |
| 468 <param name="groups_operations" value="intersect-max" /> | |
| 469 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 470 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 471 </conditional> | |
| 472 <output name="read_db" ftype="meryldb"> | |
| 473 <assert_contents> | |
| 474 <has_size value="14956" delta="300" /> | |
| 475 </assert_contents> | |
| 476 </output> | |
| 477 </test> | |
| 478 <test expect_num_outputs="1"> | |
| 479 <conditional name="operation_type"> | |
| 480 <param name="command_type" value="groups-kmers" /> | |
| 481 <param name="groups_operations" value="intersect-sum" /> | |
| 482 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 483 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 484 </conditional> | |
| 485 <output name="read_db" ftype="meryldb"> | |
| 486 <assert_contents> | |
| 487 <has_size value="14953" delta="300" /> | |
| 488 </assert_contents> | |
| 489 </output> | |
| 490 </test> | |
| 491 <!-- test 20 --> | |
| 492 <test expect_num_outputs="1"> | |
| 493 <conditional name="operation_type"> | |
| 494 <param name="command_type" value="groups-kmers" /> | |
| 495 <param name="groups_operations" value="subtract" /> | |
| 496 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 497 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 498 </conditional> | |
| 499 <output name="read_db" ftype="meryldb"> | |
| 500 <assert_contents> | |
| 501 <has_size value="23999" delta="300" /> | |
| 502 </assert_contents> | |
| 503 </output> | |
| 504 </test> | |
| 505 <test expect_num_outputs="1"> | |
| 506 <conditional name="operation_type"> | |
| 507 <param name="command_type" value="groups-kmers" /> | |
| 508 <param name="groups_operations" value="difference" /> | |
| 509 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 510 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 511 </conditional> | |
| 512 <output name="read_db" ftype="meryldb"> | |
| 513 <assert_contents> | |
| 514 <has_size value="24016" delta="300" /> | |
| 515 </assert_contents> | |
| 516 </output> | |
| 517 </test> | |
| 518 <test expect_num_outputs="1"> | |
| 519 <conditional name="operation_type"> | |
| 520 <param name="command_type" value="groups-kmers" /> | |
| 521 <param name="groups_operations" value="symmetric-difference" /> | |
| 522 <param name="input_meryldb_02" value="output_02.read-db.meryldb" ftype="meryldb" /> | |
| 523 <param name="input_meryldb_03" value="output_03.read-db.meryldb" ftype="meryldb" /> | |
| 524 </conditional> | |
| 525 <output name="read_db" ftype="meryldb"> | |
| 526 <assert_contents> | |
| 527 <has_size value="57455" delta="300" /> | |
| 528 </assert_contents> | |
| 529 </output> | |
| 530 </test> | |
| 531 <test expect_num_outputs="6"> | |
| 532 <conditional name="operation_type"> | |
| 533 <param name="command_type" value="trio-mode" /> | |
| 534 <param name="child_reads" value="child.fasta" /> | |
| 535 <param name="paternal_reads" value="paternal.fasta" /> | |
| 536 <param name="maternal_reads" value="maternal.fasta" /> | |
| 537 <conditional name="options_kmer_size"> | |
| 538 <param name="kmer_size" value="provide" /> | |
| 539 <param name="input_kmer_size" value="7" /> | |
| 540 </conditional> | |
| 541 </conditional> | |
| 542 <output name="read_db" ftype="meryldb"> | |
| 543 <assert_contents> | |
| 544 <has_size value="1573" delta="300" /> | |
| 545 </assert_contents> | |
| 546 </output> | |
| 547 <output name="read_db_hist" file="output_23.read-db.hist" /> | |
| 548 <output name="pat_db" ftype="meryldb"> | |
| 549 <assert_contents> | |
| 550 <has_size value="1779" delta="300" /> | |
| 551 </assert_contents> | |
| 552 </output> | |
| 553 <output name="pat_db_hist" file="output_23.pat.hist" /> | |
| 554 <output name="mat_db" ftype="meryldb"> | |
| 555 <assert_contents> | |
| 556 <has_size value="1569" delta="300" /> | |
| 557 </assert_contents> | |
| 558 </output> | |
| 559 <output name="mat_db_hist" file="output_23.mat.hist" /> | |
| 560 </test> | |
| 561 <test expect_num_outputs="1"> | |
| 562 <conditional name="operation_type"> | |
| 563 <param name="command_type" value="histogram-kmers" /> | |
| 564 <param name="input_meryldb_02" value="read-db.meryldb" ftype="meryldb" /> | |
| 565 </conditional> | |
| 566 <output name="read_db_hist" file="output_24.read-db.hist" /> | |
| 567 </test> | |
| 568 </tests> | |
| 569 <help><![CDATA[ | |
| 570 | |
| 571 .. class:: infomark | |
| 572 | |
| 573 **Purpose** | |
| 574 | |
| 575 Meryl is the k-mer counter. It is built into the Celera assembler and is also available as a stand-alone application. | |
| 576 Meryl uses a sorting-based approach that sorts the k-mers in lexicographical order. | |
| 577 | |
| 578 In addition of generating count-databases, meryl can perform simple operations on it. | |
| 579 | |
| 580 ----- | |
| 581 | |
| 582 .. class:: infomark | |
| 583 | |
| 584 **Basic functions** | |
| 585 | |
| 586 The functions that meryl includes are described below: | |
| 587 | |
| 588 :: | |
| 589 | |
| 590 COUNT OPERATIONS | |
| 591 - Count: count the occurrences of canonical k-mers | |
| 592 - Count-forward: count the occurreces of forward k-mers | |
| 593 - Count-reverse: count the occurreces of reverse k-mers | |
| 594 FILTERING OPERATIONS | |
| 595 - Less-than: return k-mers that occur fewer than N times in the input | |
| 596 - Greater-than: return k-mers that occur more than N times in the input | |
| 597 - Equal-to: return k-mers that occur exactly N times in the input | |
| 598 - Not-equal-to: return k-mers that do not occur exactly N times in the input | |
| 599 ARITHMETIC OPERATIONS | |
| 600 - Increase: add x to the count of each k-mer | |
| 601 - Decrease: subsctract x from the count of each k-mer | |
| 602 - Multiply: multiply the count of each k-mer by x | |
| 603 - Divide: divide the count of each k-mer by x | |
| 604 - Divide-round: divide the count of each k-mer by x and round th results | |
| 605 - Modulo: set the count of each k-mer to the remainder of the count divided by x | |
| 606 OPERATIONS ON SETS | |
| 607 - Union-min: return k-mers that occur in any input, set the count to the minimum count | |
| 608 - Union-max: return k-mers that occur in any input, set the count to the maximum count | |
| 609 - Union-sum: return k-mers that occur in any input, set the count to the sum of the counts | |
| 610 - Intersect: return k-mers that occur in all inputs, set the count to the count in the first input | |
| 611 - Intersect-min: return k-mers that occur in all inputs, set the count to the minimum count | |
| 612 - Intersect-max: return k-mers that occur in all inputs, set the count to the maximum count | |
| 613 - Intersect-sum: return k-mers that occur in all inputs, set the count to the sum of the counts | |
| 614 - Subtract: return k-mers that occur in the first input, subtracting counts from the other inputs | |
| 615 - Difference: return k-mers that occur in the first input, but none of the other inputs | |
| 616 - Symmetric-difference: return k-mers that occur in exactly one input | |
| 617 | |
| 618 ----- | |
| 619 | |
| 620 .. class:: infomark | |
| 621 | |
| 622 **Additional function: build hap-mers dbs for trios** | |
| 623 | |
| 624 In addition of the basic operations, this wrapper allows to build the hap-mers databases for trios, in accordance | |
| 625 with `merqury's recommended guidelines. <https://github.com/marbl/merqury/wiki/1.-Prepare-meryl-dbs#3-build-hap-mer-dbs-for-trios>`_ | |
| 626 | |
| 627 ]]> </help> | |
| 628 <expand macro="citations" /> | |
| 629 </tool> |