Mercurial > repos > jjohnson > cummerbund
annotate cummerbund_wrapper.xml @ 7:c2b922c9d545
Correct type for cuffdatadb in datatypes_conf.xml
| author | Jim Johnson <jj@umn.edu> |
|---|---|
| date | Wed, 10 Oct 2012 11:09:15 -0500 |
| parents | 9d25b0fd882b |
| children | b0d11fcbc3ac |
| rev | line source |
|---|---|
|
5
9d25b0fd882b
Add cuffdatadb datatype (cummeRbund SQLite DB) to cuffdata.py
Jim Johnson <jj@umn.edu>
parents:
1
diff
changeset
|
1 <tool id="cummerbund" name="cummeRbund" version="0.0.6"> |
| 0 | 2 |
| 3 <description>R package designed to aid and simplify the task of analyzing Cufflinks RNA-Seq output</description> | |
| 4 | |
| 5 <command interpreter="python"> | |
| 6 cummerbund_wrapper.py | |
| 7 --r-script ${script_file} | |
| 8 --html-report-from-directory "${output_html}" "${output_html.files_path}" | |
| 9 </command> | |
| 10 | |
| 11 <inputs> | |
| 12 <conditional name="backend_database_source"> | |
| 13 <param name="backend_database_selector" type="select" label="Will you select a backend database file from the history or do you want to build a new one using cuffdiff output?"> | |
| 14 <option value="history" selected="true">Use backend database from the history</option> | |
| 15 <option value="cuffdiff_output">Build backend database using cuffdiff output</option> | |
| 16 </param> | |
| 17 <when value="cuffdiff_output"> | |
| 18 <param format="tabular" name="isoforms_fpkm_tracking" type="data" label="Transcript FPKM tracking"/> | |
| 19 <param format="tabular" name="isoforms_exp" type="data" label="Transcript differential expression testing"/> | |
| 20 <param format="tabular" name="genes_fpkm_tracking" type="data" label="Gene FPKM tracking"/> | |
| 21 <param format="tabular" name="genes_exp" type="data" label="Gene differential expression testing"/> | |
| 22 <param format="tabular" name="tss_groups_fpkm_tracking" type="data" label="TSS groups FPKM tracking"/> | |
| 23 <param format="tabular" name="tss_groups_exp" type="data" label="TSS groups differential expression testing"/> | |
| 24 <param format="tabular" name="cds_fpkm_tracking" type="data" label="CDS FPKM tracking"/> | |
| 25 <param format="tabular" name="cds_exp_diff" type="data" label="CDS FPKM differential expression testing"/> | |
| 26 <param format="tabular" name="cds_diff" type="data" label="CDS overloading diffential expression testing"/> | |
| 27 <param format="tabular" name="promoters_diff" type="data" label="Promoters differential expression testing"/> | |
| 28 <param format="tabular" name="splicing_diff" type="data" label="Splicing differential expression testing"/> | |
| 29 <param name="rebuild" type="hidden" value="TRUE"/> | |
| 30 </when> | |
| 31 <when value="history"> | |
|
1
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
32 <param name="input_database" type="data" format="cuffdatadb" label="Select backend database (sqlite)"/> |
| 0 | 33 </when> |
| 34 </conditional> | |
| 35 <repeat name="plots" title="Plots"> | |
| 36 <param name="width" type="text" value="1280" label="The width of the image"/> | |
| 37 <param name="height" type="text" value="960" label="The height of the image"/> | |
| 38 <conditional name="plot"> | |
| 39 <param name="type" type="select" label="Plot type"> | |
| 40 <option value="density" selected="true">Density</option> | |
|
1
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
41 <option value="dispersion">Dispersion</option> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
42 <option value="fpkmSCV">Squared Coefficient of Variation</option> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
43 <option value="scatterMatrix">Scatter Matrix</option> |
| 0 | 44 <option value="boxplot">Boxplot</option> |
| 45 <option value="scatter">Scatter</option> | |
| 46 <option value="volcano">Volcano</option> | |
| 47 <option value="heatmap">Heatmap</option> | |
| 48 <option value="cluster">Cluster</option> | |
| 49 <option value="expressionplot">Expression Plot</option> | |
| 50 <option value="expressionbarplot">Expression Bar Plot</option> | |
| 51 </param> | |
| 52 <when value="density"> | |
| 53 <param name="log_mode" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Apply log10 transformation on FPKM values?"/> | |
|
1
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
54 <param name="replicates" type="boolean" truevalue="T" falsevalue="F" checked="False" label="Replicates?"/> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
55 </when> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
56 <when value="dispersion"> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
57 </when> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
58 <when value="fpkmSCV"> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
59 </when> |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
60 <when value="scatterMatrix"> |
| 0 | 61 </when> |
| 62 <when value="boxplot"> | |
| 63 <param name="log_mode" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Apply log10 transformation on FPKM values?"/> | |
| 64 </when> | |
| 65 <when value="scatter"> | |
| 66 <param name="x" type="text" label="Sample name for x axis"/> | |
| 67 <param name="y" type="text" label="Sample name for y axis"/> | |
| 68 <param name="log_mode" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Apply log10 transformation on FPKM values?"/> | |
| 69 <param name="smooth" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Add a smooth-fit regression line"/> | |
| 70 <conditional name="multiple_genes"> | |
| 71 <param name="multiple_genes_selector" type="boolean" truevalue="T" falsevalue="F" checked="False" label="Limit plot to a group of genes?"/> | |
| 72 <when value="T"> | |
| 73 <param name="features" type="select" label="Expression levels to plot?"> | |
| 74 <option value="gene" selected="true">Genes</option> | |
| 75 <option value="isoforms">Isoforms</option> | |
| 76 <option value="tss">TSS</option> | |
| 77 <option value="cds">CDS</option> | |
| 78 </param> | |
| 79 <repeat name="genes" title="Genes"> | |
| 80 <param name="gene_id" type="text" label="Gene ID"/> | |
| 81 </repeat> | |
| 82 </when> | |
| 83 <when value="F"/> | |
| 84 </conditional> | |
| 85 </when> | |
| 86 <when value="volcano"> | |
| 87 <param name="x" type="text" label="First sample name for comparison"/> | |
| 88 <param name="y" type="text" label="Second sample name for comparison"/> | |
| 89 <conditional name="multiple_genes"> | |
| 90 <param name="multiple_genes_selector" type="boolean" truevalue="T" falsevalue="F" checked="False" label="Limit plot to a group of genes?"/> | |
| 91 <when value="T"> | |
| 92 <param name="features" type="select" label="Expression levels to plot?"> | |
| 93 <option value="gene" selected="true">Genes</option> | |
| 94 <option value="isoforms">Isoforms</option> | |
| 95 <option value="tss">TSS</option> | |
| 96 <option value="cds">CDS</option> | |
| 97 </param> | |
| 98 <repeat name="genes" title="Genes"> | |
| 99 <param name="gene_id" type="text" label="Gene ID"/> | |
| 100 </repeat> | |
| 101 </when> | |
| 102 <when value="F"/> | |
| 103 </conditional> | |
| 104 </when> | |
| 105 <when value="heatmap"> | |
| 106 <param name="features" type="select" label="Expression levels to plot?"> | |
| 107 <option value="gene" selected="true">Genes</option> | |
| 108 <option value="isoforms">Isoforms</option> | |
| 109 <option value="tss">TSS</option> | |
| 110 <option value="cds">CDS</option> | |
| 111 </param> | |
| 112 <repeat name="genes" title="Genes"> | |
| 113 <param name="gene_id" type="text" label="Gene ID"/> | |
| 114 </repeat> | |
| 115 <param name="clustering" type="select" label="Cluster by"> | |
| 116 <option value="row">Row</option> | |
| 117 <option value="column">Column</option> | |
| 118 <option value="both" selected="true">Both</option> | |
| 119 <option value="none">None</option> | |
| 120 </param> | |
| 121 <param name="labcol" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="True" label="Display column labels?"/> | |
| 122 <param name="labrow" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="True" label="Display column labels?"/> | |
| 123 <param name="log_mode" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Apply log10 transformation on FPKM values?"/> | |
| 124 <param name="border" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="False" label="Draw border around plot?"/> | |
| 125 </when> | |
| 126 <when value="cluster"> | |
| 127 <param name="features" type="select" label="Expression levels to plot?"> | |
| 128 <option value="gene" selected="true">Genes</option> | |
| 129 <option value="isoforms">Isoforms</option> | |
| 130 <option value="tss">TSS</option> | |
| 131 <option value="cds">CDS</option> | |
| 132 </param> | |
| 133 <repeat name="genes" title="Genes"> | |
| 134 <param name="gene_id" type="text" label="Gene ID"/> | |
| 135 </repeat> | |
| 136 <param name="k" type="text" label="Number of pre-defined clusters to attempt to find."/> | |
| 137 <param name="iter_max" type="text" value="100" label="Max iterations"/> | |
| 138 </when> | |
| 139 <when value="expressionplot"> | |
| 140 <param name="features" type="select" label="Expression levels to plot?"> | |
| 141 <option value="gene" selected="true">Genes</option> | |
| 142 <option value="isoforms">Isoforms</option> | |
| 143 <option value="tss">TSS</option> | |
| 144 <option value="cds">CDS</option> | |
| 145 </param> | |
| 146 <param name="gene_id" type="text" label="Gene ID"/> | |
| 147 <param name="log_mode" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Apply log10 transformation on FPKM values?"/> | |
| 148 <param name="draw_summary" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="False" label="Draw a 'summary' line with mean FPKM | |
| 149 values for each condition?"/> | |
| 150 <param name="show_error_bars" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="True" label="Draw error bars?"/> | |
| 151 </when> | |
| 152 <when value="expressionbarplot"> | |
| 153 <param name="features" type="select" label="Expression levels to plot?"> | |
| 154 <option value="gene" selected="true">Genes</option> | |
| 155 <option value="isoforms">Isoforms</option> | |
| 156 <option value="tss">TSS</option> | |
| 157 <option value="cds">CDS</option> | |
| 158 </param> | |
| 159 <param name="gene_id" type="text" label="Gene ID"/> | |
| 160 <param name="log_mode" type="boolean" truevalue="T" falsevalue="F" checked="True" label="Apply log10 transformation on FPKM values?"/> | |
| 161 <param name="show_error_bars" type="boolean" truevalue="TRUE" falsevalue="FALSE" checked="True" label="Draw error bars?"/> | |
| 162 </when> | |
| 163 </conditional> | |
| 164 </repeat> | |
| 165 </inputs> | |
| 166 | |
| 167 <outputs> | |
| 168 <data format="data" name="output_database" label="${tool.name} on ${on_string}: Database File (sqlite)"> | |
| 169 <filter>backend_database_source['backend_database_selector'] == "cuffdiff_output"</filter> | |
| 170 </data> | |
| 171 <data format="html" name="output_html" label="${tool.name} on ${on_string} (HTML)"/> | |
| 172 </outputs> | |
| 173 | |
| 174 <requirements> | |
| 175 <requirement type="binary">R</requirement> | |
| 176 </requirements> | |
| 177 | |
| 178 <!--> | |
| 179 <tests> | |
| 180 <test> | |
| 181 <param name="" value=""/> | |
| 182 <output name="" file=""/> | |
| 183 </test> | |
| 184 </tests> | |
| 185 --> | |
| 186 <configfiles> | |
| 187 <configfile name="script_file"> | |
| 188 | |
| 189 ## Feature Selection ## | |
| 190 get_features <- function(myGenes, f="gene") { | |
| 191 if (f == "isoforms") | |
| 192 return(isoforms(myGenes)) | |
| 193 else if (f == "tss") | |
| 194 return(TSS(myGenes)) | |
| 195 else if (f == "cds") | |
| 196 return(CDS(myGenes)) | |
| 197 else | |
| 198 return(myGenes) | |
| 199 } | |
| 200 | |
| 201 ## Main Function ## | |
| 202 | |
| 203 ## Load cummeRbund library | |
| 204 library("cummeRbund") | |
| 205 | |
| 206 ## Initialize cuff object | |
| 207 cuff <- readCufflinks(dir = "", | |
| 208 #if $backend_database_source.backend_database_selector == "cuffdiff_output": | |
| 209 dbFile = "${output_database}", | |
| 210 geneFPKM = "${genes_fpkm_tracking}", | |
| 211 geneDiff = "${genes_exp}", | |
| 212 isoformFPKM = "${isoforms_fpkm_tracking}", | |
| 213 isoformDiff = "${isoforms_exp}", | |
| 214 TSSFPKM = "${tss_groups_fpkm_tracking}", | |
| 215 TSSDiff = "${tss_groups_exp}", | |
| 216 CDSFPKM = "${cds_fpkm_tracking}", | |
| 217 CDSExpDiff = "${cds_exp_diff}", | |
| 218 CDSDiff = "${cds_diff}", | |
| 219 promoterFile = "${promoters_diff}", | |
| 220 splicingFile = "${splicing_diff}", | |
| 221 rebuild = T) | |
| 222 #else: | |
| 223 dbFile = "${backend_database_source.input_database}", | |
| 224 rebuild = F) | |
| 225 #end if | |
| 226 | |
| 227 #for $i, $p in enumerate($plots, start=1): | |
| 228 #set $filename = "plot%02d-%s.png" % ($i, $p.plot['type']) | |
| 229 png(filename = "${filename}", width = ${p.width}, height = ${p.height}) | |
| 230 | |
| 231 ## Density plot ## | |
| 232 #if $p.plot['type'] == "density": | |
|
1
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
233 csDensity(genes(cuff),replicates=$p.plot.replicates) |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
234 devname = dev.off() |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
235 |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
236 ## Dispersion plot ## |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
237 #elif $p.plot['type'] == "dispersion": |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
238 dispersionPlot(genes(cuff)) |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
239 devname = dev.off() |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
240 |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
241 ## Squared Coefficient of Variation plot ## |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
242 #elif $p.plot['type'] == "fpkmSCV": |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
243 fpkmSCVPlot(genes(cuff)) |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
244 devname = dev.off() |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
245 |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
246 ## Scatter Matrix ## |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
247 #elif $p.plot['type'] == "scatterMatrix": |
|
6a9bc26ab8d9
Added cuffdata and cuffdatadb datatypes, cuffdiff multiselect outputs
Jim Johnson <jj@umn.edu>
parents:
0
diff
changeset
|
248 csScatterMatrix(genes(cuff)) |
| 0 | 249 devname = dev.off() |
| 250 | |
| 251 ## Boxplot ## | |
| 252 #elif $p.plot['type'] == "boxplot": | |
| 253 csBoxplot(genes(cuff)) | |
| 254 devname = dev.off() | |
| 255 | |
| 256 ## Scatter ## | |
| 257 #elif $p.plot['type'] == "scatter": | |
| 258 #if $p.plot.multiple_genes['multiple_genes_selector']: | |
| 259 myGeneIds <- c() | |
| 260 #for $g in $p.plot.multiple_genes['genes']: | |
| 261 myGeneIds <- c(myGeneIds, "$g['gene_id']") | |
| 262 #end for | |
| 263 myGenes <- getGenes(cuff, myGeneIds) | |
| 264 csScatter(get_features(myGenes, "$p.plot.multiple_genes['features']"), "${p.plot.x}", "${p.plot.y}", smooth=${p.plot.smooth}) | |
| 265 #else | |
| 266 csScatter(genes(cuff), "${p.plot.x}", "${p.plot.y}", smooth=${p.plot.smooth}) | |
| 267 #end if | |
| 268 devname = dev.off() | |
| 269 | |
| 270 ## Volcano ## | |
| 271 #elif $p.plot['type'] == "volcano": | |
| 272 #if $p.plot.multiple_genes['multiple_genes_selector']: | |
| 273 myGeneIds <- c() | |
| 274 #for $g in $p.plot.multiple_genes['genes']: | |
| 275 myGeneIds <- c(myGeneIds, "$g['gene_id']") | |
| 276 #end for | |
| 277 myGenes <- getGenes(cuff, myGeneIds) | |
| 278 csVolcano(get_features(myGenes, "$p.plot.multiple_genes['features']"), "${p.plot.x}", "${p.plot.y}") | |
| 279 #else | |
| 280 csVolcano(genes(cuff), "${p.plot.x}", "${p.plot.y}") | |
| 281 #end if | |
| 282 devname = dev.off() | |
| 283 | |
| 284 ## Heatmap ## | |
| 285 #elif $p.plot['type'] == "heatmap": | |
| 286 myGeneIds <- c() | |
| 287 #for $g in $p.plot.genes: | |
| 288 myGeneIds <- c(myGeneIds, "$g['gene_id']") | |
| 289 #end for | |
| 290 myGenes <- getGenes(cuff, myGeneIds) | |
| 291 csHeatmap(get_features(myGenes, "${p.plot.features}"), clustering="${p.plot.clustering}", labCol="${p.plot.labcol}", labRow="${p.plot.labrow}", border="${p.plot.border}") | |
| 292 devname = dev.off() | |
| 293 | |
| 294 ## Cluster ## | |
| 295 #elif $p.plot['type'] == "cluster": | |
| 296 myGeneIds <- c() | |
| 297 #for $g in $p.plot.genes: | |
| 298 myGeneIds <- c(myGeneIds, "$g['gene_id']") | |
| 299 #end for | |
| 300 myGenes <- getGenes(cuff, myGeneIds) | |
| 301 csCluster(get_features(myGenes, "${p.plot.features}"), k=${p.plot.k}, iter.max="${p.plot.iter_max}") | |
| 302 devname = dev.off() | |
| 303 | |
| 304 ## Expression Plot ## | |
| 305 #elif $p.plot['type'] == "expressionplot": | |
| 306 myGeneId <- "$p.plot.gene_id" | |
| 307 myGenes <- getGenes(cuff, myGeneId) | |
| 308 expressionPlot(get_features(myGenes, "${p.plot.features}"), drawSummary=${p.plot.draw_summary}, iter.max="${p.plot.show_error_bars}") | |
| 309 devname = dev.off() | |
| 310 | |
| 311 ## Expression Bar Plot ## | |
| 312 #elif $p.plot['type'] == "expressionbarplot": | |
| 313 myGeneId <- "$p.plot.gene_id" | |
| 314 myGenes <- getGenes(cuff, myGeneId) | |
| 315 expressionBarplot(get_features(myGenes, "${p.plot.features}"), iter.max="${p.plot.show_error_bars}") | |
| 316 devname = dev.off() | |
| 317 #end if | |
| 318 | |
| 319 #end for | |
| 320 </configfile> | |
| 321 </configfiles> | |
| 322 | |
| 323 <help> | |
| 324 This tool allows for persistent storage, access, exploration, and manipulation of Cufflinks high-throughput sequencing data. In addition, provides numerous plotting functions for commonly used visualizations. | |
| 325 </help> | |
| 326 | |
| 327 </tool> |