|
0
|
1 <tool id="ensembl_cdna_translate" name="Ensembl cDNA Translations" version="0.1.0">
|
|
|
2 <description>using Ensembl REST API</description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="package" version="0.4.10">requests-cache</requirement>
|
|
|
5 <requirement type="package" version="1.62">biopython</requirement>
|
|
|
6 </requirements>
|
|
|
7 <stdio>
|
|
|
8 <exit_code range="1:" />
|
|
|
9 </stdio>
|
|
|
10 <command><![CDATA[
|
|
|
11 #if $input:
|
|
|
12 cat '$input'
|
|
|
13 #else
|
|
|
14 python '$__tool_directory__/ensembl_cdna_translate.py'
|
|
|
15 #if $species:
|
|
|
16 --species '$species'
|
|
|
17 #end if
|
|
|
18 $transcript_raw
|
|
|
19 #if $biotypes:
|
|
|
20 --biotypes '$biotypes'
|
|
|
21 #end if
|
|
|
22 #if str($output_choice).find('transcript_bed') >= 0:
|
|
|
23 --transcripts
|
|
|
24 #if str($output_choice).find('translation') >= 0:
|
|
|
25 '-' | tee '$transcript_bed'
|
|
|
26 #else
|
|
|
27 '$transcript_bed'
|
|
|
28 #end if
|
|
|
29 #elif str($output_choice).find('translation') >= 0:
|
|
|
30 --transcripts '-'
|
|
|
31 #end if
|
|
|
32 #end if
|
|
|
33 #if str($output_choice).find('translation') >= 0:
|
|
|
34 | python '$__tool_directory__/ensembl_cdna_translate.py' -i '-'
|
|
|
35 #if $input and str($output_choice).find('transcript_bed') >= 0:
|
|
|
36 --transcripts '$transcript_bed
|
|
|
37 #end if
|
|
|
38 #if str($output_choice).find('translation_bed') >= 0:
|
|
|
39 --bed '$translation_bed'
|
|
|
40 #end if
|
|
|
41 #if str($output_choice).find('translation_fasta') >= 0:
|
|
|
42 --fasta '$translation_fasta'
|
|
|
43 #end if
|
|
|
44 #if $enzyme:
|
|
|
45 --enzyme '$enzyme'
|
|
|
46 #end if
|
|
|
47 #end if
|
|
|
48
|
|
|
49 ]]></command>
|
|
|
50 <inputs>
|
|
|
51 <param name="species" type="text" value="" label="Ensembl species" >
|
|
|
52 <help>
|
|
|
53 </help>
|
|
|
54 <option value="homo_sapiens">homo_sapiens (Human) taxon_id: 9606</option>
|
|
|
55 <option value="mus_musculus">mus_musculus (Mouse) taxon_id: 10090</option>
|
|
|
56 <option value="ailuropoda_melanoleuca">ailuropoda_melanoleuca (Panda) taxon_id: 9646</option>
|
|
|
57 <option value="anas_platyrhynchos">anas_platyrhynchos (Duck) taxon_id: 8839</option>
|
|
|
58 <option value="anolis_carolinensis">anolis_carolinensis (Anole lizard) taxon_id: 28377</option>
|
|
|
59 <option value="astyanax_mexicanus">astyanax_mexicanus (Cave fish) taxon_id: 7994</option>
|
|
|
60 <option value="bos_taurus">bos_taurus (Cow) taxon_id: 9913</option>
|
|
|
61 <option value="caenorhabditis_elegans">caenorhabditis_elegans (Caenorhabditis elegans) taxon_id: 6239</option>
|
|
|
62 <option value="callithrix_jacchus">callithrix_jacchus (Marmoset) taxon_id: 9483</option>
|
|
|
63 <option value="canis_familiaris">canis_familiaris (Dog) taxon_id: 9615</option>
|
|
|
64 <option value="carlito_syrichta">carlito_syrichta (Tarsier) taxon_id: 1868482</option>
|
|
|
65 <option value="cavia_aperea">cavia_aperea (Brazilian guinea pig) taxon_id: 37548</option>
|
|
|
66 <option value="cavia_porcellus">cavia_porcellus (Guinea Pig) taxon_id: 10141</option>
|
|
|
67 <option value="chinchilla_lanigera">chinchilla_lanigera (Long-tailed chinchilla) taxon_id: 34839</option>
|
|
|
68 <option value="chlorocebus_sabaeus">chlorocebus_sabaeus (Vervet-AGM) taxon_id: 60711</option>
|
|
|
69 <option value="choloepus_hoffmanni">choloepus_hoffmanni (Sloth) taxon_id: 9358</option>
|
|
|
70 <option value="ciona_intestinalis">ciona_intestinalis (C.intestinalis) taxon_id: 7719</option>
|
|
|
71 <option value="ciona_savignyi">ciona_savignyi (C.savignyi) taxon_id: 51511</option>
|
|
|
72 <option value="cricetulus_griseus_chok1gshd">cricetulus_griseus_chok1gshd (Chinese hamster CHOK1GS) taxon_id: 10029</option>
|
|
|
73 <option value="cricetulus_griseus_crigri">cricetulus_griseus_crigri (Chinese hamster CriGri) taxon_id: 10029</option>
|
|
|
74 <option value="danio_rerio">danio_rerio (Zebrafish) taxon_id: 7955</option>
|
|
|
75 <option value="dasypus_novemcinctus">dasypus_novemcinctus (Armadillo) taxon_id: 9361</option>
|
|
|
76 <option value="dipodomys_ordii">dipodomys_ordii (Kangaroo rat) taxon_id: 10020</option>
|
|
|
77 <option value="drosophila_melanogaster">drosophila_melanogaster (Fruitfly) taxon_id: 7227</option>
|
|
|
78 <option value="echinops_telfairi">echinops_telfairi (Lesser hedgehog tenrec) taxon_id: 9371</option>
|
|
|
79 <option value="equus_caballus">equus_caballus (Horse) taxon_id: 9796</option>
|
|
|
80 <option value="erinaceus_europaeus">erinaceus_europaeus (Hedgehog) taxon_id: 9365</option>
|
|
|
81 <option value="felis_catus">felis_catus (Cat) taxon_id: 9685</option>
|
|
|
82 <option value="ficedula_albicollis">ficedula_albicollis (Flycatcher) taxon_id: 59894</option>
|
|
|
83 <option value="fukomys_damarensis">fukomys_damarensis (Damara mole rat) taxon_id: 885580</option>
|
|
|
84 <option value="gadus_morhua">gadus_morhua (Cod) taxon_id: 8049</option>
|
|
|
85 <option value="gallus_gallus">gallus_gallus (Chicken) taxon_id: 9031</option>
|
|
|
86 <option value="gasterosteus_aculeatus">gasterosteus_aculeatus (Stickleback) taxon_id: 69293</option>
|
|
|
87 <option value="gorilla_gorilla">gorilla_gorilla (Gorilla) taxon_id: 9595</option>
|
|
|
88 <option value="heterocephalus_glaber_female">heterocephalus_glaber_female (Naked mole-rat female) taxon_id: 10181</option>
|
|
|
89 <option value="heterocephalus_glaber_male">heterocephalus_glaber_male (Naked mole-rat male) taxon_id: 10181</option>
|
|
|
90 <option value="ictidomys_tridecemlineatus">ictidomys_tridecemlineatus (Squirrel) taxon_id: 43179</option>
|
|
|
91 <option value="jaculus_jaculus">jaculus_jaculus (Lesser Egyptian jerboa) taxon_id: 51337</option>
|
|
|
92 <option value="latimeria_chalumnae">latimeria_chalumnae (Coelacanth) taxon_id: 7897</option>
|
|
|
93 <option value="lepisosteus_oculatus">lepisosteus_oculatus (Spotted gar) taxon_id: 7918</option>
|
|
|
94 <option value="loxodonta_africana">loxodonta_africana (Elephant) taxon_id: 9785</option>
|
|
|
95 <option value="macaca_mulatta">macaca_mulatta (Macaque) taxon_id: 9544</option>
|
|
|
96 <option value="meleagris_gallopavo">meleagris_gallopavo (Turkey) taxon_id: 9103</option>
|
|
|
97 <option value="mesocricetus_auratus">mesocricetus_auratus (Golden Hamster) taxon_id: 10036</option>
|
|
|
98 <option value="microcebus_murinus">microcebus_murinus (Mouse Lemur) taxon_id: 30608</option>
|
|
|
99 <option value="microtus_ochrogaster">microtus_ochrogaster (Prairie vole) taxon_id: 79684</option>
|
|
|
100 <option value="monodelphis_domestica">monodelphis_domestica (Opossum) taxon_id: 13616</option>
|
|
|
101 <option value="mus_caroli">mus_caroli (Ryukyu mouse) taxon_id: 10089</option>
|
|
|
102 <option value="mus_musculus_129s1svimj">mus_musculus_129s1svimj (Mouse 129S1/SvImJ) taxon_id: 10090</option>
|
|
|
103 <option value="mus_musculus_aj">mus_musculus_aj (Mouse A/J) taxon_id: 10090</option>
|
|
|
104 <option value="mus_musculus_akrj">mus_musculus_akrj (Mouse AKR/J) taxon_id: 10090</option>
|
|
|
105 <option value="mus_musculus_balbcj">mus_musculus_balbcj (Mouse BALB/cJ) taxon_id: 10090</option>
|
|
|
106 <option value="mus_musculus_c3hhej">mus_musculus_c3hhej (Mouse C3H/HeJ) taxon_id: 10090</option>
|
|
|
107 <option value="mus_musculus_c57bl6nj">mus_musculus_c57bl6nj (Mouse C57BL/6NJ) taxon_id: 10090</option>
|
|
|
108 <option value="mus_musculus_casteij">mus_musculus_casteij (Mouse CAST/EiJ) taxon_id: 10091</option>
|
|
|
109 <option value="mus_musculus_cbaj">mus_musculus_cbaj (Mouse CBA/J) taxon_id: 10090</option>
|
|
|
110 <option value="mus_musculus_dba2j">mus_musculus_dba2j (Mouse DBA/2J) taxon_id: 10090</option>
|
|
|
111 <option value="mus_musculus_fvbnj">mus_musculus_fvbnj (Mouse FVB/NJ) taxon_id: 10090</option>
|
|
|
112 <option value="mus_musculus_lpj">mus_musculus_lpj (Mouse LP/J) taxon_id: 10090</option>
|
|
|
113 <option value="mus_musculus_nodshiltj">mus_musculus_nodshiltj (Mouse NOD/ShiLtJ) taxon_id: 10090</option>
|
|
|
114 <option value="mus_musculus_nzohlltj">mus_musculus_nzohlltj (Mouse NZO/HlLtJ) taxon_id: 10090</option>
|
|
|
115 <option value="mus_musculus_pwkphj">mus_musculus_pwkphj (Mouse PWK/PhJ) taxon_id: 39442</option>
|
|
|
116 <option value="mus_musculus_wsbeij">mus_musculus_wsbeij (Mouse WSB/EiJ) taxon_id: 10092</option>
|
|
|
117 <option value="mus_pahari">mus_pahari (Shrew mouse) taxon_id: 10093</option>
|
|
|
118 <option value="mus_spretus_spreteij">mus_spretus_spreteij (Algerian mouse) taxon_id: 10096</option>
|
|
|
119 <option value="mustela_putorius_furo">mustela_putorius_furo (Ferret) taxon_id: 9669</option>
|
|
|
120 <option value="myotis_lucifugus">myotis_lucifugus (Microbat) taxon_id: 59463</option>
|
|
|
121 <option value="nannospalax_galili">nannospalax_galili (Upper Galilee mountains blind mole rat) taxon_id: 1026970</option>
|
|
|
122 <option value="nomascus_leucogenys">nomascus_leucogenys (Gibbon) taxon_id: 61853</option>
|
|
|
123 <option value="notamacropus_eugenii">notamacropus_eugenii (Wallaby) taxon_id: 9315</option>
|
|
|
124 <option value="ochotona_princeps">ochotona_princeps (Pika) taxon_id: 9978</option>
|
|
|
125 <option value="octodon_degus">octodon_degus (Degu) taxon_id: 10160</option>
|
|
|
126 <option value="oreochromis_niloticus">oreochromis_niloticus (Tilapia) taxon_id: 8128</option>
|
|
|
127 <option value="ornithorhynchus_anatinus">ornithorhynchus_anatinus (Platypus) taxon_id: 9258</option>
|
|
|
128 <option value="oryctolagus_cuniculus">oryctolagus_cuniculus (Rabbit) taxon_id: 9986</option>
|
|
|
129 <option value="oryzias_latipes">oryzias_latipes (Medaka) taxon_id: 8090</option>
|
|
|
130 <option value="otolemur_garnettii">otolemur_garnettii (Bushbaby) taxon_id: 30611</option>
|
|
|
131 <option value="ovis_aries">ovis_aries (Sheep) taxon_id: 9940</option>
|
|
|
132 <option value="pan_troglodytes">pan_troglodytes (Chimpanzee) taxon_id: 9598</option>
|
|
|
133 <option value="papio_anubis">papio_anubis (Olive baboon) taxon_id: 9555</option>
|
|
|
134 <option value="pelodiscus_sinensis">pelodiscus_sinensis (Chinese softshell turtle) taxon_id: 13735</option>
|
|
|
135 <option value="peromyscus_maniculatus_bairdii">peromyscus_maniculatus_bairdii (Northern American deer mouse) taxon_id: 230844</option>
|
|
|
136 <option value="petromyzon_marinus">petromyzon_marinus (Lamprey) taxon_id: 7757</option>
|
|
|
137 <option value="poecilia_formosa">poecilia_formosa (Amazon molly) taxon_id: 48698</option>
|
|
|
138 <option value="pongo_abelii">pongo_abelii (Orangutan) taxon_id: 9601</option>
|
|
|
139 <option value="procavia_capensis">procavia_capensis (Hyrax) taxon_id: 9813</option>
|
|
|
140 <option value="pteropus_vampyrus">pteropus_vampyrus (Megabat) taxon_id: 132908</option>
|
|
|
141 <option value="rattus_norvegicus">rattus_norvegicus (Rat) taxon_id: 10116</option>
|
|
|
142 <option value="saccharomyces_cerevisiae">saccharomyces_cerevisiae (Saccharomyces cerevisiae) taxon_id: 4932</option>
|
|
|
143 <option value="sarcophilus_harrisii">sarcophilus_harrisii (Tasmanian devil) taxon_id: 9305</option>
|
|
|
144 <option value="sorex_araneus">sorex_araneus (Shrew) taxon_id: 42254</option>
|
|
|
145 <option value="sus_scrofa">sus_scrofa (Pig) taxon_id: 9823</option>
|
|
|
146 <option value="taeniopygia_guttata">taeniopygia_guttata (Zebra Finch) taxon_id: 59729</option>
|
|
|
147 <option value="takifugu_rubripes">takifugu_rubripes (Fugu) taxon_id: 31033</option>
|
|
|
148 <option value="tetraodon_nigroviridis">tetraodon_nigroviridis (Tetraodon) taxon_id: 99883</option>
|
|
|
149 <option value="tupaia_belangeri">tupaia_belangeri (Tree Shrew) taxon_id: 37347</option>
|
|
|
150 <option value="tursiops_truncatus">tursiops_truncatus (Dolphin) taxon_id: 9739</option>
|
|
|
151 <option value="vicugna_pacos">vicugna_pacos (Alpaca) taxon_id: 30538</option>
|
|
|
152 <option value="xenopus_tropicalis">xenopus_tropicalis (Xenopus) taxon_id: 8364</option>
|
|
|
153 <option value="xiphophorus_maculatus">xiphophorus_maculatus (Platyfish) taxon_id: 8083</option>
|
|
|
154
|
|
|
155 </param>
|
|
|
156 <param name="biotypes" type="text" value="" optional="true" label="Restrict to these biotypes" >
|
|
|
157 <help><![CDATA[
|
|
|
158 Example biotypes:
|
|
|
159 protein_coding, non_coding, pseudogene, nonsense_mediated_decay, non_stop_decay,
|
|
|
160 translated_processed_pseudogene, transcribed_processed_pseudogene, transcribed_unitary_pseudogene, transcribed_unprocessed_pseudogene,
|
|
|
161 polymorphic_pseudogene, processed_pseudogene, unprocessed_pseudogene, unitary_pseudogene, processed_transcript,
|
|
|
162 retained_intron, ccds_gene, sense_overlapping, sense_intronic, cdna_update, antisense,
|
|
|
163 LRG_gene, IG_C_gene, IG_D_gene, IG_J_gene, IG_LV_gene IG_V_gene, TR_C_gene, TR_D_gene, TR_J_gene, TR_V_gene,
|
|
|
164 IG_pseudogene, IG_C_pseudogene, IG_D_pseudogene, IG_J_pseudogene, IG_V_pseudogene, TR_J_pseudogene, TR_V_pseudogene, TEC,
|
|
|
165 ribozyme, RNase_P_RNA, guide_RNA, macro_lncRNA, bidirectional_promoter_lncRNA, 3prime_overlapping_ncRNA, antisense_RNA, vaultRNA, Y_RNA, SRP_RNA, RNase_MRP_RNA, IG_C_pseudogene, lncRNA, lincRNA, miRNA, snRNA, sRNA, telomerase_RNA, Mt_tRNA, Mt_rRNA, scaRNA, misc_RNA, rRNA, tRNA, scRNA, snoRNA, other
|
|
|
166 ]]></help>
|
|
|
167 </param>
|
|
|
168 <param name="input" type="data" format="bed" optional="true" label="A BED file with 12 columns, thickStart and thickEnd define protein coding region"/>
|
|
|
169 <param name="translate_all" type="boolean" truevalue="--all" falsevalue="" checked="false"
|
|
|
170 label="Report all translations (Default is non reference protein sequences)"/>
|
|
|
171 <param name="transcript_raw" type="boolean" truevalue="--raw" falsevalue="" checked="true"
|
|
|
172 label="Keep extra columns from ensembl BED"/>
|
|
|
173 <param name="output_choice" type="select" multiple="true" display="checkboxes" label="Outputs">
|
|
|
174 <option value="transcript_bed">transcripts.bed</option>
|
|
|
175 <option value="translation_bed">translation.bed</option>
|
|
|
176 <option value="translation_fasta">translation.fasta</option>
|
|
|
177 </param>
|
|
|
178 <param name="min_length" type="integer" value="7" min="1" label="Minimum length of protein translation to report"/>
|
|
|
179 <param name="enzyme" type="select" optional="true" label="Digest enzyme"
|
|
|
180 help="Remove frags that are in a reference protein">
|
|
|
181 <option value="trypsin">trypsin: ([KR](?=[^P]))|((?<=W)K(?=P))|((?<=M)R(?=P))</option>
|
|
|
182 </param>
|
|
|
183 </inputs>
|
|
|
184 <outputs>
|
|
|
185 <data name="transcript_bed" format="bed" label="Ensembl $species transcripts.bed">
|
|
|
186 <filter>'transcript_bed' in output_choice</filter>
|
|
|
187 </data>
|
|
|
188 <data name="translation_bed" format="bed" label="Ensembl $species translation.bed">
|
|
|
189 <filter>'translation_bed' in output_choice</filter>
|
|
|
190 </data>
|
|
|
191 <data name="translation_fasta" format="fasta" label="Ensembl $species translation.fasta">
|
|
|
192 <filter>'translation_fasta' in output_choice</filter>
|
|
|
193 </data>
|
|
|
194 </outputs>
|
|
|
195 <tests>
|
|
|
196 <test>
|
|
|
197 <param name="species" value="human"/>
|
|
|
198 <param name="input" value="human_transcripts.bed" ftype="bed"/>
|
|
|
199 <param name="output_choice" value="translation_bed,translation_fasta"/>
|
|
|
200 <output name="translation_bed">
|
|
|
201 <assert_contents>
|
|
|
202 <has_text text="ENST00000641515" />
|
|
|
203 </assert_contents>
|
|
|
204 </output>
|
|
|
205 <output name="translation_fasta">
|
|
|
206 <assert_contents>
|
|
|
207 <has_text text=">ENST00000641515" />
|
|
|
208 </assert_contents>
|
|
|
209 </output>
|
|
|
210 </test>
|
|
|
211 </tests>
|
|
|
212 <help><![CDATA[
|
|
|
213 usage: ensembl_cdna_translate.py [-h] [-s SPECIES] [-i INPUT] [-t TRANSCRIPTS]
|
|
|
214 [-r] [-f FASTA] [-b BED] [-m MIN_LENGTH] [-a]
|
|
|
215 [-v] [-d]
|
|
|
216
|
|
|
217 Retrieve Ensembl cDNAs and three frame translate
|
|
|
218
|
|
|
219 optional arguments:
|
|
|
220 -h, --help show this help message and exit
|
|
|
221 -s SPECIES, --species SPECIES
|
|
|
222 Ensembl Species to retrieve
|
|
|
223 -i INPUT, --input INPUT
|
|
|
224 Use this bed instead of retrieving cDNA from ensembl
|
|
|
225 (-) for stdin
|
|
|
226 -t TRANSCRIPTS, --transcripts TRANSCRIPTS
|
|
|
227 Path to output cDNA transcripts.bed (-) for stdout
|
|
|
228 -r, --raw Report transcript exacty as returned from Ensembl
|
|
|
229 -f FASTA, --fasta FASTA
|
|
|
230 Path to output translations.fasta
|
|
|
231 -b BED, --bed BED Path to output translations.bed
|
|
|
232 -m MIN_LENGTH, --min_length MIN_LENGTH
|
|
|
233 Minimum length of protein translation to report
|
|
|
234 -a, --all Report all translations (Default is non reference
|
|
|
235 protein sequences)
|
|
|
236 -v, --verbose Verbose
|
|
|
237 -d, --debug Debug
|
|
|
238
|
|
|
239 Esmebl REST API returns a 20 BED format with these additional columns::
|
|
|
240
|
|
|
241 second_name, cds_start_status, cds_end_status, exon_frames, type, gene_name, second_gene_name, gene_type
|
|
|
242
|
|
|
243 ]]></help>
|
|
|
244 <citations>
|
|
|
245 <citation type="doi">10.1093/bioinformatics/btu613</citation>
|
|
|
246 <citation type="doi">10.1093/nar/gku1010</citation>
|
|
|
247 </citations>
|
|
|
248 </tool>
|
