Commit 9baa6300 authored by Remi  PLANEL's avatar Remi PLANEL
Browse files

Use custom gff parser : legiolist-gff

parent 0805cf85
......@@ -3,16 +3,9 @@
<property name="common.os.prefix" value="common"/>
<sources>
<source name="LegioParis-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila str. Paris chromosome sequence"/>
<property name="fasta.taxonId" value="297246"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.includes" value="LegioParis_10-05-2012_LegioList.fasta"/>
</source>
<!-- First gff and then fasta if you want to have the sequence available -->
<source name="LegioParis-gff" type="gff">
<source name="LegioParis-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="297246"/>
<property name="gff3.seqDataSourceName" value="LegioList LegioParis"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -20,29 +13,35 @@
<property name="gff3.dataSetTitle" value="LegioList Legionella pneumophila str. Paris genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/LegioParis"/>
</source>
<source name="LegioParis-uniprot" type="uniprot">
<property name="uniprot.organisms" value="297246"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/uniprot"/>
<property name="creatego" value="true"/>
<property name="creategenes" value="true"/>
<property name="allowduplicates" value="false"/>
<property name="loadfragments" value="false"/>
<property name="loadtrembl" value="true"/>
</source>
<source name="Lfallonii-chromosome-fasta" type="fasta">
<source name="LegioParis-chromosome-fasta" type="fasta">
<property name="fasta.taxonId" value="297246"/>
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella fallonii chromosome sequence"/>
<property name="fasta.taxonId" value="96230"/>
<property name="fasta.includes" value="Lfallonii.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila str. Paris chromosome sequence"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
<property name="fasta.includes" value="LegioParis_10-05-2012_LegioList.fasta"/>
</source>
<source name="LegioParis-ncbi-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="297246"/>
<property name="gff3.seqDataSourceName" value="NCBI LegioParis"/>
<property name="gff3.dataSourceName" value="NCBI"/>
<property name="gff3.seqClsName" value="Chromosome"/>
<property name="gff3.dataSetTitle" value="NCBI Legionella pneumophila str. Paris annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/ncbi/LegioParis"/>
</source>
<source name="LegioParis-ncbi-fasta" type="fasta">
<property name="fasta.taxonId" value="297246"/>
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="NCBI"/>
<property name="fasta.dataSetTitle" value="NCBI Legionella pneumophila str. Paris chromosome sequence"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/ncbi"/>
<property name="fasta.includes" value="GCF_000048645.1_ASM4864v1_genomic.fna"/>
</source>
<source name="Lfallonii-gff" type="gff">
<source name="Lfallonii-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="96230"/>
<property name="gff3.seqDataSourceName" value="LegioList Lfalloni"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -50,29 +49,18 @@
<property name="gff3.dataSetTitle" value="Legionella fallonii genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/Lfallonii"/>
</source>
<source name="Lfallonii-uniprot" type="uniprot">
<property name="uniprot.organisms" value="96230"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/uniprot"/>
<property name="creatego" value="true"/>
<property name="creategenes" value="true"/>
<property name="allowduplicates" value="false"/>
<property name="loadfragments" value="false"/>
<property name="loadtrembl" value="true"/>
</source>
<source name="Lhackeliae-chromosome-fasta" type="fasta">
<source name="Lfallonii-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella hackeliae chromosome sequence"/>
<property name="fasta.taxonId" value="449"/>
<property name="fasta.includes" value="Lhackeliae.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella fallonii chromosome sequence"/>
<property name="fasta.taxonId" value="96230"/>
<property name="fasta.includes" value="Lfallonii.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="Lhackeliae-gff" type="gff">
<source name="Lhackeliae-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="449"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella hackeliae"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -80,21 +68,18 @@
<property name="gff3.dataSetTitle" value="Legionella hackeliae genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/Lhackeliae"/>
</source>
<source name="Lmicadei-chromosome-fasta" type="fasta">
<source name="Lhackeliae-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella micdadei chromosome sequence"/>
<property name="fasta.taxonId" value="451"/>
<property name="fasta.includes" value="Lmicadei.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella hackeliae chromosome sequence"/>
<property name="fasta.taxonId" value="449"/>
<property name="fasta.includes" value="Lhackeliae.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="Lmicadei-gff" type="gff">
<source name="Lmicadei-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="451"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella micdadei"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -102,20 +87,18 @@
<property name="gff3.dataSetTitle" value="Legionella micdadei genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/Lmicadei"/>
</source>
<source name="Long_Mage-chromosome-fasta" type="fasta">
<source name="Lmicadei-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella longbeachae NSW150 chromosome sequence"/>
<property name="fasta.taxonId" value="661367"/>
<property name="fasta.includes" value="Long_Mage.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella micdadei chromosome sequence"/>
<property name="fasta.taxonId" value="451"/>
<property name="fasta.includes" value="Lmicadei.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="Long_Mage-gff" type="gff">
<source name="Long_Mage-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="661367"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella longbeachae NSW150"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -123,20 +106,19 @@
<property name="gff3.dataSetTitle" value="Legionella longbeachae NSW150 genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/Long_Mage"/>
</source>
<source name="lorraine-chromosome-fasta" type="fasta">
<source name="Long_Mage-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila subsp. pneumophila str. Lorraine chromosome sequence"/>
<property name="fasta.taxonId" value="1046632"/>
<property name="fasta.includes" value="Lorraine.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella longbeachae NSW150 chromosome sequence"/>
<property name="fasta.taxonId" value="661367"/>
<property name="fasta.includes" value="Long_Mage.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="Lorraine-gff" type="gff">
<source name="Lorraine-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="1046632"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella pneumophila subsp. pneumophila str. Lorraine"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -144,19 +126,20 @@
<property name="gff3.dataSetTitle" value="Legionella pneumophila subsp. pneumophila str. Lorraine genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/Lorraine"/>
</source>
<source name="LpAlcoy-chromosome-fasta" type="fasta">
<source name="lorraine-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila 2300/99 Alcoy chromosome sequence"/>
<property name="fasta.taxonId" value="423212"/>
<property name="fasta.includes" value="LpAlcoy.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila subsp. pneumophila str. Lorraine chromosome sequence"/>
<property name="fasta.taxonId" value="1046632"/>
<property name="fasta.includes" value="Lorraine.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="LpAlcoy-gff" type="gff">
<source name="LpAlcoy-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="423212"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella pneumophila 2300/99 Alcoy"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -164,19 +147,19 @@
<property name="gff3.dataSetTitle" value="Legionella pneumophila 2300/99 Alcoy genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/LpAlcoy"/>
</source>
<source name="LpCorby-chromosome-fasta" type="fasta">
<source name="LpAlcoy-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila str. Corby chromosome sequence"/>
<property name="fasta.taxonId" value="400673"/>
<property name="fasta.includes" value="LpCorby.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila 2300/99 Alcoy chromosome sequence"/>
<property name="fasta.taxonId" value="423212"/>
<property name="fasta.includes" value="LpAlcoy.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="LpCorby-gff" type="gff">
<source name="LpCorby-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="400673"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella pneumophila str. Corby"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -184,21 +167,21 @@
<property name="gff3.dataSetTitle" value="Legionella pneumophila str. Corby genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/LpCorby"/>
</source>
<source name="LpCorby-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila str. Corby chromosome sequence"/>
<property name="fasta.taxonId" value="400673"/>
<property name="fasta.includes" value="LpCorby.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="LpLens-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila str. Lens chromosome sequence"/>
<property name="fasta.taxonId" value="297245"/>
<property name="fasta.includes" value="LpLens.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
</source>
<source name="LpLens-gff" type="gff">
<source name="LpLens-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="297245"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella pneumophila str. Lens"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -206,18 +189,18 @@
<property name="gff3.dataSetTitle" value="Legionella pneumophila str. Lens genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/LpLens"/>
</source>
<source name="LpPhila-chromosome-fasta" type="fasta">
<source name="LpLens-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila subsp. pneumophila str. Philadelphia 1 chromosome sequence"/>
<property name="fasta.taxonId" value="272624"/>
<property name="fasta.includes" value="LpPhila.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila str. Lens chromosome sequence"/>
<property name="fasta.taxonId" value="297245"/>
<property name="fasta.includes" value="LpLens.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="LpPhila-gff" type="gff">
<source name="LpPhila-gff" type="legiolist-gff">
<property name="gff3.taxonId" value="272624"/>
<property name="gff3.seqDataSourceName" value="LegioList Legionella pneumophila subsp. pneumophila str. Philadelphia 1"/>
<property name="gff3.dataSourceName" value="LegioList"/>
......@@ -225,7 +208,24 @@
<property name="gff3.dataSetTitle" value="Legionella pneumophila subsp. pneumophila str. Philadelphia 1 genome annotation"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/gff/legiolist-gff/LpPhila"/>
</source>
<source name="LpPhila-chromosome-fasta" type="fasta">
<property name="fasta.className" value="org.intermine.model.bio.Chromosome"/>
<property name="fasta.dataSourceName" value="LegioList"/>
<property name="fasta.dataSetTitle" value="LegioList Legionella pneumophila subsp. pneumophila str. Philadelphia 1 chromosome sequence"/>
<property name="fasta.taxonId" value="272624"/>
<property name="fasta.includes" value="LpPhila.fasta"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/fasta/legiolist"/>
</source>
<source name="uniprot" type="uniprot">
<property name="uniprot.organisms" value="272624 297246 297245 400673 423212 1046632 661367 451 449 96230"/>
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/uniprot"/>
<property name="creatego" value="true"/>
<property name="creategenes" value="true"/>
<property name="allowduplicates" value="false"/>
<property name="loadfragments" value="true"/>
<property name="loadtrembl" value="true"/>
</source>
<source name="legiolist-homology" type="legiolist-homology">
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/"/>
......@@ -243,12 +243,22 @@
<property name="licence" value="https://creativecommons.org/licenses/by/4.0/"/>
</source>
<source name="go" type="go">
<property name="src.data.file" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/go/go-basic.obo"/>
</source>
<source name="pubmed-gene" type="pubmed-gene">
<property name="src.data.dir" location="/home/rplanel/projects/legiolist-intermine/data/legiolist/genomes/publication"/>
<property name="pubmed.organisms" value="272624 297246 297245 400673 423212 1046632 661367 451 449 96230"/>
<property name="src.data.dir.includes" value="gene2pubmed"/>
</source>
<source name="update-publications" type="update-publications" dump="true">
<property name="src.data.file" location="publications.xml"/>
<!-- <property name="loadFullRecord" value="true"/> -->
</source>
</sources>
<post-processing>
......@@ -256,6 +266,7 @@
<post-process name="transfer-sequences"/>
<post-process name="create-chromosome-locations-and-lengths"/>
<post-process name="do-sources" />
<post-process name="populate-child-features"/>
<post-process name="create-attribute-indexes" />
<post-process name="create-search-index" />
<post-process name="summarise-objectstore" />
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment