source: trunk/flymine/project.xml @ 29409

Revision 29409, 37.7 KB checked in by julie, 25 hours ago (diff)

replace old rnai source with new one

Line 
1<project type="bio">
2  <property name="target.model" value="genomic"/>
3  <property name="source.location" location="../bio/sources/"/>
4  <property name="source.location" location="../bio/sources/flymine"/>
5  <property name="common.os.prefix" value="common"/>
6  <property name="intermine.properties.file" value="flymine.properties"/>
7  <property name="default.intermine.properties.file" location="../default.intermine.integrate.properties"/>
8  <sources>
9
10    <source name="chado-db-flybase-dmel" type="chado-db" dump="true">
11      <property name="source.db.name" value="flybase"/>
12      <property name="organisms" value="7227"/>
13      <property name="dataSourceName" value="FlyBase"/>
14      <property name="processors"
15                value="org.intermine.bio.dataconversion.FlyBaseProcessor org.intermine.bio.dataconversion.StockProcessor"/>
16      <property name="converter.class"
17                value="org.intermine.bio.dataconversion.ChadoDBConverter"/>
18    </source>
19    <source name="chado-db-flybase-dpse" type="chado-db">
20      <property name="source.db.name" value="flybase"/>
21      <property name="organisms" value="7237"/>
22      <property name="dataSourceName" value="FlyBase"/>
23      <property name="processors"
24                value="org.intermine.bio.dataconversion.FlyBaseProcessor"/>
25      <property name="converter.class"
26                value="org.intermine.bio.dataconversion.ChadoDBConverter"/>
27    </source>
28    <source name="chado-db-flybase-others" type="chado-db" dump="true">
29      <property name="source.db.name" value="flybase"/>
30      <property name="organisms" value="Dana Dere Dgri Dmoj Dper Dsec Dsim Dvir Dwil Dyak"/>
31      <property name="dataSourceName" value="FlyBase"/>
32      <property name="processors"
33                value="org.intermine.bio.dataconversion.FlyBaseProcessor"/>
34      <property name="converter.class"
35                value="org.intermine.bio.dataconversion.ChadoDBConverter"/>
36    </source>
37
38    <source name="wormbase-identifiers" type="wormbase-identifiers">
39      <property name="src.data.dir" location="/micklem/data/worm-identifiers/current"/>
40    </source>
41    <source name="ensembl-hgnc" type="ensembl-hgnc">
42     <property name="src.data.dir" location="/micklem/data/ensembl-ids/current"/>
43    </source> 
44   <source name="ncbi-gene" type="ncbi-gene">
45     <property name="src.data.dir" location="/micklem/data/ncbi/current/"/>
46     <property name="organisms" value="9606"/>
47    </source>
48   <source name="mgi-identifiers" type="mgi-identifiers">
49      <property name="src.data.dir" location="/micklem/data/mgi-identifiers/current"/>
50    </source>
51   <source name="rgd-identifiers" type="rgd-identifiers">
52      <property name="src.data.dir" location="/micklem/data/rgd-identifiers/current"/>
53    </source>
54
55    <!-- uniprot must be processed immediately after chado -->
56    <source name="uniprot" type="uniprot" dump="true">
57      <property name="uniprot.organisms" value="7227 6239 7165 9606 10090 10116 46245 7230 7240 7244 7245 7217 7220 7222 7234 7260 7238 559292"/>
58      <property name="src.data.dir" location="/micklem/data/uniprot/current"/>
59      <property name="createinterpro" value="true"/>
60    </source>
61
62    <source name="uniprot-fasta" type="fasta">
63      <property name="fasta.taxonId" value="7227 6239 7165 4932 9606 10090 10116 46245 7230 7240 7244 7245 7217 7220 7222 7234 7260 7238"/>
64      <property name="fasta.className" value="org.intermine.model.bio.Protein"/>
65      <property name="fasta.classAttribute" value="primaryAccession"/>
66      <property name="fasta.dataSetTitle" value="UniProt data set"/>
67      <property name="fasta.dataSourceName" value="UniProt"/>
68      <property name="src.data.dir" location="/micklem/data/uniprot/current"/>
69      <property name="fasta.includes" value="uniprot_sprot_varsplic.fasta"/>
70      <property name="fasta.sequenceType" value="protein" />
71      <property name="fasta.loaderClassName"
72                value="org.intermine.bio.dataconversion.UniProtFastaLoaderTask"/>
73    </source>
74     <source name="drosophila-homology" type="drosophila-homology">
75      <property name="src.data.dir" location="/micklem/data/flybase/homology/current"/>
76      <property name="src.data.dir.includes" value="*.tsv"/>
77    </source>
78    <source name="anopheles-identifiers" type="anopheles-identifiers">
79      <property name="src.data.dir" location="/micklem/data/anopheles-identifiers/current"/>
80    </source>
81    <source name="uniprot-keywords" type="uniprot-keywords">
82      <property name="src.data.dir" location="/micklem/data/uniprot/current"/>
83      <property name="src.data.dir.includes" value="keywlist.xml"/>
84    </source>
85    <source name="rnai" type="rnai">
86      <property name="src.data.dir" location="/micklem/data/flymine/rnai/current"/>
87    </source>
88    <source name="arbeitman-items-xml" type="arbeitman-items-xml">
89      <property name="src.data.file" location="/micklem/data/flymine/mage/arbeitman/arbeitman-tgt-items.xml"/>
90    </source>
91    <source name="bdgp-clone" type="bdgp-clone">
92      <property name="src.data.dir" location="/micklem/data/flymine/bdgp-clones"/>
93      <property name="src.data.dir.includes" value="dgc2flybase.txt"/>
94    </source>
95
96    <source name="long-oligo" type="long-oligo" dump="true">
97      <property name="gff3.taxonId" value="7227"/>
98      <property name="gff3.seqClsName" value="MRNA"/>
99      <property name="src.data.dir" location="/micklem/data/flymine/long-oligo"/>
100    </source>
101
102    <source name="ensembl" type="ensembl">
103     <property name="src.data.dir" location="/micklem/data/ensembl/current"/>
104     <property name="src.data.dir.includes" value="7165.xml"/>
105    </source>
106
107    <source name="interpro" type="interpro">
108      <property name="src.data.dir" location="/micklem/data/interpro/current"/>
109    </source>
110
111    <source name="omim" type="omim">
112      <property name="src.data.dir" location="/micklem/data/metabolic/omim/test"/>
113    </source>
114    <source name="kegg-pathway" type="kegg-pathway">
115      <property name="src.data.dir" location="/micklem/data/kegg/current"/>
116      <property name="kegg.organisms" value="7227"/>
117    </source>
118<!--
119    <source name="kegg-orthologues" type="kegg-orthologues">
120      <property name="src.data.dir" location="/micklem/data/kegg/orthologues"/>
121      <property name="kegg.organisms" value="7227 9606 4932 6239 7955 10090 10116"/>
122    </source>
123-->
124    <source name="flyatlas" type="flyatlas">
125      <property name="src.data.dir" location="/micklem/data/flyatlas/current"/>
126    </source>
127    <source name="anoph-expr" type="anoph-expr">
128      <property name="src.data.dir" location="/micklem/data/flymine/anoph-expr"/>
129      <property name="src.data.dir.includes" value="E-TABM-186-processed-data-1343527732.txt"/>
130      <property name="geneFile" value="/micklem/data/flymine/anoph-expr/reporter_gene_mappingFile.txt"/>
131    </source>
132
133    <source name="treefam" type="treefam">
134      <property name="src.data.dir" location="/micklem/data/treefam/current"/>
135      <property name="src.data.dir.includes" value="ortholog.txt.table"/>
136      <property name="geneFile" value="/micklem/data/treefam/current/genes.txt.table"/>
137      <property name="treefam.organisms" value="7227 7165"/>
138      <property name="treefam.homologues" value="9606 10090 10116 7955 6239 4932"/>
139    </source>
140
141    <source name="ensembl-compara" type="ensembl-compara">
142      <property name="src.data.dir" location="/micklem/data/ensembl-compara/current"/>
143      <property name="ensemblcompara.organisms" value="7227"/>
144      <property name="ensemblcompara.homologues" value="6239 9606 10090 10116"/>
145    </source>
146    <source name="anoest" type="anoest">
147      <property name="dataSourceName" value="VectorBase"/>
148      <property name="dataSetTitle" value="AnoEST clusters"/>
149      <property name="source.db.name" value="anoest"/>
150    </source>
151    <source name="pdb" type="pdb">
152      <property name="src.data.dir" location="/micklem/data/pdb/current"/>
153      <property name="pdb.organisms" value="7227"/>
154    </source>
155    <source name="psi-mi-ontology" type="psi-mi-ontology">
156      <property name="src.data.file" location="/micklem/data/psi/ontology/current/psi-mi.obo"/>
157    </source>
158    <source name="so" type="so">
159      <property name="src.data.file" location="../../bio/sources/so/so.obo"/>
160    </source>
161    <source name="go" type="go" dump="true">
162      <property name="src.data.file" location="/micklem/data/go-annotation/current/gene_ontology.obo"/>
163    </source>
164<!--
165    <source name="uberon" type="uberon">
166      <property name="src.data.file" location="/micklem/data/old_shared_data/uberon/uberon.obo"/>
167    </source>
168-->
169    <source name="go-annotation" type="ontology-annotation">
170      <property name="ontologyfile" location="/micklem/data/go-annotation/current/gene_ontology.obo"/>
171      <property name="src.data.dir" location="/micklem/data/go-annotation/flymine"/>
172      <property name="ontologyPrefix" value="GO"/>
173    </source>
174   
175    <source name="flybase-expression" type="flybase-expression">
176      <property name="src.data.dir" location="/micklem/data/modMine_flyRNAseq"/>
177      <property name="src.data.dir.includes" value="HTD_modENCODE_BinData_2010-10-05.txt"/>
178      <property name="flybaseExpressionLevelsFile" location="/micklem/data/modMine_flyRNAseq/BinLOOKUP_modENCODE_FlyAtlas.txt"/>
179    </source>
180
181    <source name="flyreg" type="flyreg">
182      <property name="gff3.taxonId" value="7227"/>
183      <property name="gff3.seqClsName" value="Chromosome"/>
184      <property name="src.data.dir" location="/micklem/data/redfly/current/tfbs"/>
185      <property name="src.data.dir.includes" value="tfbs_dump.gff"/>
186    </source>
187    <source name="redfly" type="redfly">
188      <property name="gff3.taxonId" value="7227"/>
189      <property name="gff3.seqClsName" value="Chromosome"/>
190      <property name="src.data.dir" location="/micklem/data/redfly/current/crm"/>
191      <property name="src.data.dir.includes" value="crm_dump.gff"/>
192    </source>
193
194    <source name="fly-fish" type="fly-fish">
195      <property name="src.data.dir" location="/micklem/data/flymine/flyfish"/>
196      <property name="src.data.dir.includes" value="annotation_matrix_filter_3.csv"/>
197    </source>
198    <source name="flybase-dmel-gene-fasta" type="fasta">
199      <property name="fasta.taxonId" value="7227"/>
200      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila melanogaster"/>
201      <property name="fasta.dataSourceName" value="FlyBase"/>
202      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
203      <property name="fasta.classAttribute" value="primaryIdentifier"/>
204      <property name="fasta.includes" value="dmel-all-gene-*.fasta"/>
205      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
206    </source>
207    <source name="flybase-dpse-gene-fasta" type="fasta">
208      <property name="fasta.taxonId" value="7237"/>
209      <property name="fasta.dataSetTitle"
210                value="FlyBase fasta data set for Drosophila pseudoobscura pseudoobscura"/>
211      <property name="fasta.dataSourceName" value="FlyBase"/>
212      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
213      <property name="fasta.classAttribute" value="primaryIdentifier"/>
214      <property name="fasta.includes" value="dpse-all-gene-*.fasta"/>
215      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
216    </source>
217    <source name="flybase-dana-gene-fasta" type="fasta">
218      <property name="fasta.taxonId" value="7217"/>
219      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila ananassae"/>
220      <property name="fasta.dataSourceName" value="FlyBase"/>
221      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
222      <property name="fasta.classAttribute" value="primaryIdentifier"/>
223      <property name="fasta.includes" value="dana-all-gene-*.fasta"/>
224      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
225    </source>
226    <source name="flybase-dere-gene-fasta" type="fasta">
227      <property name="fasta.taxonId" value="7220"/>
228      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila erecta"/>
229      <property name="fasta.dataSourceName" value="FlyBase"/>
230      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
231      <property name="fasta.classAttribute" value="primaryIdentifier"/>
232      <property name="fasta.includes" value="dere-all-gene-*.fasta"/>
233      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
234    </source>
235    <source name="flybase-dgri-gene-fasta" type="fasta">
236      <property name="fasta.taxonId" value="7222"/>
237      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila grimshawi"/>
238      <property name="fasta.dataSourceName" value="FlyBase"/>
239      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
240      <property name="fasta.classAttribute" value="primaryIdentifier"/>
241      <property name="fasta.includes" value="dgri-all-gene-*.fasta"/>
242      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
243    </source>
244    <source name="flybase-dmoj-gene-fasta" type="fasta">
245      <property name="fasta.taxonId" value="7230"/>
246      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila mojavensis"/>
247      <property name="fasta.dataSourceName" value="FlyBase"/>
248      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
249      <property name="fasta.classAttribute" value="primaryIdentifier"/>
250      <property name="fasta.includes" value="dmoj-all-gene-*.fasta"/>
251      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
252    </source>
253    <source name="flybase-dper-gene-fasta" type="fasta">
254      <property name="fasta.taxonId" value="7234"/>
255      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila persimilis"/>
256      <property name="fasta.dataSourceName" value="FlyBase"/>
257      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
258      <property name="fasta.classAttribute" value="primaryIdentifier"/>
259      <property name="fasta.includes" value="dper-all-gene-*.fasta"/>
260      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
261    </source>
262    <source name="flybase-dsec-gene-fasta" type="fasta">
263      <property name="fasta.taxonId" value="7238"/>
264      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila sechellia"/>
265      <property name="fasta.dataSourceName" value="FlyBase"/>
266      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
267      <property name="fasta.classAttribute" value="primaryIdentifier"/>
268      <property name="fasta.includes" value="dsec-all-gene-*.fasta"/>
269      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
270    </source>
271    <source name="flybase-dsim-gene-fasta" type="fasta">
272      <property name="fasta.taxonId" value="7240"/>
273      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila simulans"/>
274      <property name="fasta.dataSourceName" value="FlyBase"/>
275      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
276      <property name="fasta.classAttribute" value="primaryIdentifier"/>
277      <property name="fasta.includes" value="dsim-all-gene-*.fasta"/>
278      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
279    </source>
280    <source name="flybase-dvir-gene-fasta" type="fasta">
281      <property name="fasta.taxonId" value="7244"/>
282      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila virilis"/>
283      <property name="fasta.dataSourceName" value="FlyBase"/>
284      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
285      <property name="fasta.classAttribute" value="primaryIdentifier"/>
286      <property name="fasta.includes" value="dvir-all-gene-*.fasta"/>
287      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
288    </source>
289    <source name="flybase-dwil-gene-fasta" type="fasta">
290      <property name="fasta.taxonId" value="7260"/>
291      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila willistoni"/>
292      <property name="fasta.dataSourceName" value="FlyBase"/>
293      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
294      <property name="fasta.classAttribute" value="primaryIdentifier"/>
295      <property name="fasta.includes" value="dwil-all-gene-*.fasta"/>
296      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
297    </source>
298    <source name="flybase-dyak-gene-fasta" type="fasta">
299      <property name="fasta.taxonId" value="7245"/>
300      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila yakuba"/>
301      <property name="fasta.dataSourceName" value="FlyBase"/>
302      <property name="fasta.className" value="org.intermine.model.bio.Gene"/>
303      <property name="fasta.classAttribute" value="primaryIdentifier"/>
304      <property name="fasta.includes" value="dyak-all-gene-*.fasta"/>
305      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
306    </source>
307    <source name="flybase-dmel-cds-fasta" type="fasta">
308      <property name="fasta.loaderClassName"
309                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
310      <property name="fasta.taxonId" value="7227"/>
311      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
312      <property name="fasta.classAttribute" value="primaryIdentifier"/>
313      <property name="fasta.includes" value="dmel-all-CDS-*.fasta"/>
314      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila melanogaster"/>
315      <property name="fasta.dataSourceName" value="FlyBase"/>
316      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
317    </source>
318    <source name="flybase-dpse-cds-fasta" type="fasta">
319      <property name="fasta.loaderClassName"
320                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
321      <property name="fasta.taxonId" value="7237"/>
322      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
323      <property name="fasta.classAttribute" value="primaryIdentifier"/>
324      <property name="fasta.includes" value="dpse-all-CDS-*.fasta"/>
325      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila pseudoobscura pseudoobscura"/>
326      <property name="fasta.dataSourceName" value="FlyBase"/>
327      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
328    </source>
329    <source name="flybase-dana-cds-fasta" type="fasta">
330      <property name="fasta.loaderClassName"
331                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
332      <property name="fasta.taxonId" value="7217"/>
333      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
334      <property name="fasta.classAttribute" value="primaryIdentifier"/>
335      <property name="fasta.includes" value="dana-all-CDS-*.fasta"/>
336      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila ananassae"/>
337      <property name="fasta.dataSourceName" value="FlyBase"/>
338      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
339    </source>
340    <source name="flybase-dere-cds-fasta" type="fasta">
341      <property name="fasta.loaderClassName"
342                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
343      <property name="fasta.taxonId" value="7220"/>
344      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
345      <property name="fasta.classAttribute" value="primaryIdentifier"/>
346      <property name="fasta.includes" value="dere-all-CDS-*.fasta"/>
347      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila erecta"/>
348      <property name="fasta.dataSourceName" value="FlyBase"/>
349      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
350    </source>
351    <source name="flybase-dgri-cds-fasta" type="fasta">
352      <property name="fasta.loaderClassName"
353                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
354      <property name="fasta.taxonId" value="7222"/>
355      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
356      <property name="fasta.classAttribute" value="primaryIdentifier"/>
357      <property name="fasta.includes" value="dgri-all-CDS-*.fasta"/>
358      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila grimshawi"/>
359      <property name="fasta.dataSourceName" value="FlyBase"/>
360      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
361    </source>
362    <source name="flybase-dmoj-cds-fasta" type="fasta">
363      <property name="fasta.loaderClassName"
364                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
365      <property name="fasta.taxonId" value="7230"/>
366      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
367      <property name="fasta.classAttribute" value="primaryIdentifier"/>
368      <property name="fasta.includes" value="dmoj-all-CDS-*.fasta"/>
369      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila mojavensis"/>
370      <property name="fasta.dataSourceName" value="FlyBase"/>
371      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
372    </source>
373    <source name="flybase-dper-cds-fasta" type="fasta">
374      <property name="fasta.loaderClassName"
375                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
376      <property name="fasta.taxonId" value="7234"/>
377      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
378      <property name="fasta.classAttribute" value="primaryIdentifier"/>
379      <property name="fasta.includes" value="dper-all-CDS-*.fasta"/>
380      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila persimilis"/>
381      <property name="fasta.dataSourceName" value="FlyBase"/>
382      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
383    </source>
384    <source name="flybase-dsec-cds-fasta" type="fasta">
385      <property name="fasta.loaderClassName"
386                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
387      <property name="fasta.taxonId" value="7238"/>
388      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
389      <property name="fasta.classAttribute" value="primaryIdentifier"/>
390      <property name="fasta.includes" value="dsec-all-CDS-*.fasta"/>
391      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila sechellia"/>
392      <property name="fasta.dataSourceName" value="FlyBase"/>
393      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
394    </source>
395    <source name="flybase-dsim-cds-fasta" type="fasta">
396      <property name="fasta.loaderClassName"
397                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
398      <property name="fasta.taxonId" value="7240"/>
399      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
400      <property name="fasta.classAttribute" value="primaryIdentifier"/>
401      <property name="fasta.includes" value="dsim-all-CDS-*.fasta"/>
402      <property name="fasta.dataSetTitle" value="FlyBase fasta data set for Drosophila simulans"/>
403      <property name="fasta.dataSourceName" value="FlyBase"/>
404      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
405    </source>
406    <source name="flybase-dvir-cds-fasta" type="fasta">
407      <property name="fasta.loaderClassName"
408                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
409      <property name="fasta.taxonId" value="7244"/>
410      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
411      <property name="fasta.classAttribute" value="primaryIdentifier"/>
412      <property name="fasta.includes" value="dvir-all-CDS-*.fasta"/>
413      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila virilis"/>
414      <property name="fasta.dataSourceName" value="FlyBase"/>
415      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
416    </source>
417    <source name="flybase-dwil-cds-fasta" type="fasta">
418      <property name="fasta.loaderClassName"
419                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
420      <property name="fasta.taxonId" value="7260"/>
421      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
422      <property name="fasta.classAttribute" value="primaryIdentifier"/>
423      <property name="fasta.includes" value="dwil-all-CDS-*.fasta"/>
424      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila willistoni"/>
425      <property name="fasta.dataSourceName" value="FlyBase"/>
426      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
427    </source>
428    <source name="flybase-dyak-cds-fasta" type="fasta">
429      <property name="fasta.loaderClassName"
430                value="org.intermine.bio.dataconversion.FlyBaseCDSFastaLoaderTask"/>
431      <property name="fasta.taxonId" value="7245"/>
432      <property name="fasta.className" value="org.intermine.model.bio.CDS"/>
433      <property name="fasta.classAttribute" value="primaryIdentifier"/>
434      <property name="fasta.includes" value="dyak-all-CDS-*.fasta"/>
435      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila yakuba"/>
436      <property name="fasta.dataSourceName" value="FlyBase"/>
437      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
438    </source>
439    <source name="flybase-dmel-5prime-utr-fasta" type="fasta">
440      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
441      <property name="fasta.taxonId" value="7227"/>
442      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
443      <property name="fasta.classAttribute" value="primaryIdentifier"/>
444      <property name="fasta.includes" value="dmel-all-five_prime_UTR-*.fasta"/>
445      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila melanogaster"/>
446      <property name="fasta.dataSourceName" value="FlyBase"/>
447      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
448    </source>
449    <source name="flybase-dmel-3prime-utr-fasta" type="fasta">
450      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
451      <property name="fasta.taxonId" value="7227"/>
452      <property name="fasta.className" value="org.intermine.model.bio.ThreePrimeUTR"/>
453      <property name="fasta.classAttribute" value="primaryIdentifier"/>
454      <property name="fasta.includes" value="dmel-all-three_prime_UTR-*.fasta"/>
455      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila melanogaster"/>
456      <property name="fasta.dataSourceName" value="FlyBase"/>
457      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
458    </source>
459   
460   
461    <source name="flybase-dpse-5prime-utr-fasta" type="fasta">
462      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
463      <property name="fasta.taxonId" value="7237"/>
464      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
465      <property name="fasta.classAttribute" value="primaryIdentifier"/>
466      <property name="fasta.includes" value="dpse-all-five_prime_UTR-*.fasta"/>
467      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila pseudoobscura pseudoobscura"/>
468      <property name="fasta.dataSourceName" value="FlyBase"/>
469      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
470    </source>
471    <source name="flybase-dpse-3prime-utr-fasta" type="fasta">
472      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
473      <property name="fasta.taxonId" value="7237"/>
474      <property name="fasta.className" value="org.intermine.model.bio.ThreePrimeUTR"/>
475      <property name="fasta.classAttribute" value="primaryIdentifier"/>
476      <property name="fasta.includes" value="dpse-all-three_prime_UTR-*.fasta"/>
477      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila pseudoobscura pseudoobscura"/>
478      <property name="fasta.dataSourceName" value="FlyBase"/>
479      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
480    </source>
481    <source name="flybase-dgri-5prime-utr-fasta" type="fasta">
482      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
483      <property name="fasta.taxonId" value="7222"/>
484      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
485      <property name="fasta.classAttribute" value="primaryIdentifier"/>
486      <property name="fasta.includes" value="dgri-all-five_prime_UTR-*.fasta"/>
487      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila grimshawi"/>
488      <property name="fasta.dataSourceName" value="FlyBase"/>
489      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
490    </source>
491    <source name="flybase-dper-5prime-utr-fasta" type="fasta">
492      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
493      <property name="fasta.taxonId" value="7234"/>
494      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
495      <property name="fasta.classAttribute" value="primaryIdentifier"/>
496      <property name="fasta.includes" value="dper-all-five_prime_UTR-*.fasta"/>
497      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila persimilis"/>
498      <property name="fasta.dataSourceName" value="FlyBase"/>
499      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
500    </source>
501    <source name="flybase-dsec-5prime-utr-fasta" type="fasta">
502      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
503      <property name="fasta.taxonId" value="7238"/>
504      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
505      <property name="fasta.classAttribute" value="primaryIdentifier"/>
506      <property name="fasta.includes" value="dsec-all-five_prime_UTR-*.fasta"/>
507      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila sechellia"/>
508      <property name="fasta.dataSourceName" value="FlyBase"/>
509      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
510    </source>
511    <source name="flybase-dsim-5prime-utr-fasta" type="fasta">
512      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
513      <property name="fasta.taxonId" value="7240"/>
514      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
515      <property name="fasta.classAttribute" value="primaryIdentifier"/>
516      <property name="fasta.includes" value="dsim-all-five_prime_UTR-*.fasta"/>
517      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila simulans"/>
518      <property name="fasta.dataSourceName" value="FlyBase"/>
519      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
520    </source>
521    <source name="flybase-dyak-5prime-utr-fasta" type="fasta">
522      <property name="fasta.loaderClassName" value="org.intermine.bio.dataconversion.FlyBaseUTRFastaLoaderTask"/>
523      <property name="fasta.taxonId" value="7245"/>
524      <property name="fasta.className" value="org.intermine.model.bio.FivePrimeUTR"/>
525      <property name="fasta.classAttribute" value="primaryIdentifier"/>
526      <property name="fasta.includes" value="dyak-all-five_prime_UTR-*.fasta"/>
527      <property name="fasta.dataSetTitle" value="FlyBase data set for Drosophila yakuba"/>
528      <property name="fasta.dataSourceName" value="FlyBase"/>
529      <property name="src.data.dir" location="/micklem/data/flybase/fasta/current"/>
530    </source>
531    <source name="pubmed-gene" type="pubmed-gene" dump="true">
532      <property name="src.data.dir" location="/micklem/data/pubmed/current"/>
533      <property name="pubmed.organisms" value="7227 7165 6239 7240 7244 7245 7217 7220 7222 7234 7238 46245"/>
534      <property name="src.data.dir.includes" value="gene2pubmed"/>
535      <property name="infoFile" location="/micklem/data/pubmed/current/gene_info"/>
536    </source>
537
538    <source name="interpro-go" type="interpro-go">
539      <property name="src.data.dir" location="/micklem/data/interpro/ontology/current"/>
540    </source>
541
542<!-- text files, didn't see a gain from BioPAX
543    <source name="reactome" type="reactome">
544      <property name="src.data.dir" location="/micklem/data/old_shared_data/reactome/current"/>
545      <property name="reactome.organisms" value="7227 4932 6239"/>
546    </source>
547-->
548
549    <source name="reactome" type="biopax">
550      <property name="src.data.dir" location="/micklem/data/reactome/current"/>
551      <property name="biopax.organisms" value="7227"/>
552      <property name="biopax.datasourcename" value="Reactome"/>
553      <property name="biopax.datasetname" value="Reactome data set"/>
554      <property name="biopax.curated" value="false"/>
555    </source>
556
557    <source name="reactome-curated" type="biopax">
558      <property name="src.data.dir" location="/micklem/data/reactome/curated"/>
559      <property name="biopax.organisms" value="7227"/>
560      <property name="biopax.datasourcename" value="Reactome"/>
561      <property name="biopax.datasetname" value="FlyReactome data set"/>
562      <property name="biopax.curated" value="true"/>
563    </source>
564
565    <source name="bdgp-insitu" type="bdgp-insitu">
566      <property name="src.data.dir" location="/micklem/data/flymine/bdgp-insitu/current"/>
567    </source>
568    <source name="fly-anatomy-ontology" type="fly-anatomy-ontology">
569      <property name="src.data.file" location="/micklem/data/ontologies/fly-anatomy/current/fly_anatomy.obo"/>
570    </source>
571    <source name="fly-development-ontology" type="fly-development-ontology">
572      <property name="src.data.file" location="/micklem/data/ontologies/fly-development/current/fly_development.obo"/>
573    </source>
574    <source name="fly-misc-cvterms" type="fly-misc-cvterms">
575      <property name="src.data.file" location="/micklem/data/ontologies/fly-cv/current/flybase_controlled_vocabulary.obo"/>
576    </source>
577
578    <source name="drosdel-gff" type="drosdel-gff">
579      <property name="gff3.taxonId" value="7227"/>
580      <property name="gff3.seqClsName" value="Chromosome"/>
581      <property name="src.data.dir"
582                location="/micklem/data/flymine/drosdel/2008-03-19/chromosome_structure_variations/"/>
583    </source>
584
585    <source name="miranda" type="miranda">
586      <property name="gff3.seqClsName" value="Chromosome"/>
587      <property name="gff3.taxonId" value="7227"/>
588      <property name="src.data.dir" location="/micklem/data/flymine/mirna/miranda/current"/>
589      <property name="src.data.dir.includes" value="v5.gff.drosophila_melanogaster.gff"/>
590    </source>
591    <source name="psi-intact" type="psi" dump="true">
592      <property name="src.data.dir" location="/micklem/data/psi/intact/current"/>
593      <property name="intact.organisms" value="7227 6239 46245 7240 7244 7245"/>
594    </source>
595
596    <source name="biogrid" type="biogrid">
597      <property name="src.data.dir" location="/micklem/data/biogrid/current"/>
598      <property name="src.data.dir.includes" value="*psi25.xml"/>
599      <property name="biogrid.organisms" value="7227 6239"/>
600    </source>
601
602    <source name="affy-probes" type="affy-probes">
603      <property name="src.data.dir" location="/micklem/data/flymine/affy-probes/current"/>
604      <property name="src.data.dir.includes" value="*.txt"/>
605    </source>
606
607    <source name="entrez-organism" type="entrez-organism">
608      <!-- this source finds the taxon ids in the objectstore, retrieves
609           species name and genus from Entrez into an InterMine format XML
610           file, loads that file into a target items DB, then into the final
611           objectstore -->
612      <property name="src.data.file" location="build/organisms.xml"/>
613    </source>
614    <source name="flymine-static" type="intermine-items-xml-file">
615      <property name="src.data.file" location="datasets.xml"/>
616    </source>
617    <source name="update-publications" type="update-publications" dump="true">
618      <!-- this source finds the publication ids (pubmed ids) in the
619           objectstore, retrieves the details of the publication from Entrez
620           into an InterMine format XML file, loads that file into a target
621           items DB, then into the final objectstore -->
622      <property name="src.data.file" location="build/publications.xml"/>
623    </source>
624  </sources>
625
626  <post-processing>
627
628
629    <post-process name="create-utr-references"/>
630    <post-process name="create-references" />
631    <post-process name="make-spanning-locations"/>
632    <post-process name="create-chromosome-locations-and-lengths"/>
633    <post-process name="create-intergenic-region-features"/>
634    <post-process name="transfer-sequences"/>
635    <post-process name="create-overlap-relations-flymine" dump="true"/>
636    <post-process name="do-sources" />
637    <post-process name="create-intron-features">
638      <property name="organisms" value="7165"/>
639    </post-process>
640    <post-process name="create-bioseg-location-index"/>
641    <post-process name="create-overlap-view" />
642    <post-process name="create-attribute-indexes"/>
643    <post-process name="summarise-objectstore"/>
644    <post-process name="create-autocomplete-index"/>
645    <post-process name="create-search-index"/>
646
647    <!-- grant-db-production/
648    TODO: Not really post processing but just has to be done sometime...
649  <target name="grant-db-production" depends="prepare" - ->
650    <!- - grant 'public' user select permission on production db - ->
651    <taskdef name="grant" classname="org.intermine.task.GrantTask">
652      <classpath refid="models.class.path"/>
653    </taskdef>
654
655    <grant database="db.production" user="PUBLIC" perm="SELECT"/>
656
657  </target>
658    -->
659  </post-processing>
660
661</project>
662
Note: See TracBrowser for help on using the repository browser.