Splitting paint

metadata/datasets.schema.yaml

@@ -103,6 +103,66 @@ mapping:
             type: str
             required: false
             unique: false
+          "paint":
+            type: map
+            required: false
+            mapping:
+              "id":
+                type: str
+                required: true
+                unique: true
+              "label":
+                type: str
+                required: true
+                unique: true
+              "description":
+                type: str
+                required: true
+                unique: true
+              "url":
+                type: str
+                required: true
+                unique: true
+              "type":
+                type: str
+                required: true
+                unique: false
+              "dataset":
+                type: str
+                required: true
+                unique: false
+              "submitter":
+                type: str
+                required: true
+                unique: false
+              "compression":
+                type: str
+                required: false
+                unique: false
+              "source":
+                type: str
+                required: false
+                unique: false
+              "species_code":
+                type: str
+                required: false
+                unique: false
+              "merges_into":
+                type: str
+                required: false
+                unique: false
+              "entity_type":
+                type: str
+                required: false
+                unique: false
+              "aggregates":
+                type: str
+                required: false
+                unique: false
+              "status":
+                type: str
+                required: false
+                unique: false
           "taxa":
             type: seq
             required: false

metadata/datasets/aspgd.yaml

@@ -7,40 +7,39 @@ project_url: "http://aspergillusgenomes.org/"
 funding_source: "NIAID US NIH"
 email_report: "[email protected]"
 datasets:
- -    
-   id: aspgd.gaf
-   label: "aspgd gaf file"
-   description: "gaf file for aspgd from Aspergillus Genome Database"
-   url: http://geneontology.org/gene-associations/gene_association.aspgd.gz
-   type: gaf
-   dataset: aspgd
-   submitter: aspgd
-   compression: gzip
-   source: http://geneontology.org/gene-associations/submission/gene_association.aspgd.gz
-   entity_type: 
-   status: active
-   species_code: Aspergillus
-   taxa:
-    - NCBITaxon:5052
-    - NCBITaxon:1033177
-    - NCBITaxon:1073089
-    - NCBITaxon:1137211
-    - NCBITaxon:1160497
-    - NCBITaxon:331117
-    - NCBITaxon:332952
-    - NCBITaxon:341663
-    - NCBITaxon:344612
-    - NCBITaxon:41063
-    - NCBITaxon:451804
-    - NCBITaxon:46472
-    - NCBITaxon:5061
-    - NCBITaxon:5062
-    - NCBITaxon:602072
-    - NCBITaxon:690307
-    - NCBITaxon:746128
-    - NCBITaxon:75750
-    - NCBITaxon:767769
-    - NCBITaxon:767770
-    - NCBITaxon:162425
-    - NCBITaxon:5085
-
+  -
+    id: aspgd.gaf
+    label: "aspgd gaf file"
+    description: "gaf file for aspgd from Aspergillus Genome Database"
+    url: http://geneontology.org/gene-associations/gene_association.aspgd.gz
+    type: gaf
+    dataset: aspgd
+    submitter: aspgd
+    compression: gzip
+    source: http://geneontology.org/gene-associations/submission/gene_association.aspgd.gz
+    entity_type:
+    status: active
+    species_code: Aspergillus
+    taxa:
+      - NCBITaxon:5052
+      - NCBITaxon:1033177
+      - NCBITaxon:1073089
+      - NCBITaxon:1137211
+      - NCBITaxon:1160497
+      - NCBITaxon:331117
+      - NCBITaxon:332952
+      - NCBITaxon:341663
+      - NCBITaxon:344612
+      - NCBITaxon:41063
+      - NCBITaxon:451804
+      - NCBITaxon:46472
+      - NCBITaxon:5061
+      - NCBITaxon:5062
+      - NCBITaxon:602072
+      - NCBITaxon:690307
+      - NCBITaxon:746128
+      - NCBITaxon:75750
+      - NCBITaxon:767769
+      - NCBITaxon:767770
+      - NCBITaxon:162425
+      - NCBITaxon:5085

metadata/datasets/cgd.yaml

@@ -7,29 +7,40 @@ project_url: "http://www.candidagenome.org/"
 funding_source: "NIDCR at US NIH, grant number 1-R01-DE015873-01"
 email_report: "[email protected]"
 datasets:
- -    
-   id: cgd.gaf
-   label: "cgd gaf file"
-   description: "gaf file for cgd from Candida Genome Database"
-   url: http://geneontology.org/gene-associations/gene_association.cgd.gz
-   type: gaf
-   dataset: cgd
-   submitter: cgd
-   compression: gzip
-   source: http://geneontology.org/gene-associations/submission/gene_association.cgd.gz
-   entity_type: 
-   status: active
-   species_code: 
-   taxa:
-    - NCBITaxon:5476
-    - NCBITaxon:5478
-    - NCBITaxon:5480
-    - NCBITaxon:1136231
-    - NCBITaxon:284592
-    - NCBITaxon:294746
-    - NCBITaxon:294747
-    - NCBITaxon:294748
-    - NCBITaxon:306902
-    - NCBITaxon:379508
-    - NCBITaxon:42374
-
+  -
+    id: cgd.gaf
+    label: "cgd gaf file"
+    description: "gaf file for cgd from Candida Genome Database"
+    url: http://geneontology.org/gene-associations/gene_association.cgd.gz
+    type: gaf
+    dataset: cgd
+    submitter: cgd
+    compression: gzip
+    source: http://geneontology.org/gene-associations/submission/gene_association.cgd.gz
+    entity_type:
+    status: active
+    paint:
+      id: paint_cgd.gaf
+      label: "paint_cgd gaf file"
+      description: "gaf file for paint_cgd from PAINT"
+      url: http://geneontology.org/gene-associations/gene_association.paint_cgd.gaf.gz
+      type: gaf
+      dataset: paint_cgd
+      submitter: paint
+      compression: gzip
+      source: ftp://ftp.pantherdb.org/downloads/paint/presubmission/gene_association.paint_cgd.gaf.gz
+      entity_type:
+      status: active
+    species_code:
+    taxa:
+      - NCBITaxon:5476
+      - NCBITaxon:5478
+      - NCBITaxon:5480
+      - NCBITaxon:1136231
+      - NCBITaxon:284592
+      - NCBITaxon:294746
+      - NCBITaxon:294747
+      - NCBITaxon:294748
+      - NCBITaxon:306902
+      - NCBITaxon:379508
+      - NCBITaxon:42374

metadata/datasets/dictybase.yaml

@@ -7,22 +7,33 @@ project_url: "http://dictybase.org"
 funding_source: "NIH (NIGMS and NHGRI)"
 email_report: "[email protected]"
 datasets:
- -    
-   id: dictybase.gaf
-   label: "dictybase gaf file"
-   description: "gaf file for dictybase from dictyBase"
-   url: http://geneontology.org/gene-associations/gene_association.dictyBase.gz
-   type: gaf
-   dataset: dictybase
-   submitter: dictybase
-   compression: gzip
-   source: http://geneontology.org/gene-associations/submission/gene_association.dictyBase.gz
-   entity_type: 
-   status: active
-   species_code: Ddis
-   taxa:
-    - NCBITaxon:352472
-    - NCBITaxon:366501
-    - NCBITaxon:44689
-    - NCBITaxon:5782
-
+  -
+    id: dictybase.gaf
+    label: "dictybase gaf file"
+    description: "gaf file for dictybase from dictyBase"
+    url: http://geneontology.org/gene-associations/gene_association.dictyBase.gz
+    type: gaf
+    dataset: dictybase
+    submitter: dictybase
+    compression: gzip
+    source: http://geneontology.org/gene-associations/submission/gene_association.dictyBase.gz
+    entity_type:
+    status: active
+    paint:
+      id: paint_dictybase.gaf
+      label: "paint_dictybase gaf file"
+      description: "gaf file for paint_dictybase from PAINT"
+      url: http://geneontology.org/gene-associations/gene_association.paint_dictyBase.gaf.gz
+      type: gaf
+      dataset: paint_dictybase
+      submitter: paint
+      compression: gzip
+      source: ftp://ftp.pantherdb.org/downloads/paint/presubmission/gene_association.paint_dictyBase.gaf.gz
+      entity_type:
+      status: active
+    species_code: Ddis
+    taxa:
+      - NCBITaxon:352472
+      - NCBITaxon:366501
+      - NCBITaxon:44689
+      - NCBITaxon:5782

metadata/datasets/ecocyc.yaml

@@ -7,19 +7,30 @@ project_url: "http://ecoliwiki.net"
 funding_source: "GM077678, NIGMS U24GM088849"
 email_report: "[email protected]"
 datasets:
- -    
-   id: ecocyc.gaf
-   label: "ecocyc gaf file"
-   description: "gaf file for ecocyc from Encyclopedia of E. coli metabolism"
-   url: http://geneontology.org/gene-associations/gene_association.ecocyc.gz
-   type: gaf
-   dataset: ecocyc
-   submitter: ecocyc
-   compression: gzip
-   source: http://geneontology.org/gene-associations/submission/gene_association.ecocyc.gz
-   entity_type: 
-   status: active
-   species_code: 
-   taxa:
-    - NCBITaxon:83333
-
+  -
+    id: ecocyc.gaf
+    label: "ecocyc gaf file"
+    description: "gaf file for ecocyc from Encyclopedia of E. coli metabolism"
+    url: http://geneontology.org/gene-associations/gene_association.ecocyc.gz
+    type: gaf
+    dataset: ecocyc
+    submitter: ecocyc
+    compression: gzip
+    source: http://geneontology.org/gene-associations/submission/gene_association.ecocyc.gz
+    entity_type:
+    status: active
+    paint:
+      id: paint_ecocyc.gaf
+      label: "paint_ecocyc gaf file"
+      description: "gaf file for paint_ecocyc from PAINT"
+      url: http://geneontology.org/gene-associations/gene_association.paint_ecocyc.gaf.gz
+      type: gaf
+      dataset: paint_ecocyc
+      submitter: paint
+      compression: gzip
+      source: ftp://ftp.pantherdb.org/downloads/paint/presubmission/gene_association.paint_ecocyc.gaf.gz
+      entity_type:
+      status: active
+    species_code:
+    taxa:
+      - NCBITaxon:83333

metadata/datasets/fb.yaml

@@ -7,19 +7,30 @@ project_url: "http://flybase.org"
 funding_source: "MRC at UK MRC, grant number G1000968 "
 email_report: "[email protected]"
 datasets:
- -    
-   id: fb.gaf
-   label: "fb gaf file"
-   description: "gaf file for fb from FlyBase"
-   url: http://geneontology.org/gene-associations/gene_association.fb.gz
-   type: gaf
-   dataset: fb
-   submitter: fb
-   compression: gzip
-   source: http://geneontology.org/gene-associations/submission/gene_association.fb.gz
-   entity_type: 
-   status: active
-   species_code: Dmel
-   taxa:
-    - NCBITaxon:7227
-
+  -
+    id: fb.gaf
+    label: "fb gaf file"
+    description: "gaf file for fb from FlyBase"
+    url: http://geneontology.org/gene-associations/gene_association.fb.gz
+    type: gaf
+    dataset: fb
+    submitter: fb
+    compression: gzip
+    source: http://geneontology.org/gene-associations/submission/gene_association.fb.gz
+    entity_type:
+    status: active
+    paint:
+      id: paint_fb.gaf
+      label: "paint_fb gaf file"
+      description: "gaf file for paint_fb from PAINT"
+      url: http://geneontology.org/gene-associations/gene_association.paint_fb.gaf.gz
+      type: gaf
+      dataset: paint_fb
+      submitter: paint
+      compression: gzip
+      source: ftp://ftp.pantherdb.org/downloads/paint/presubmission/gene_association.paint_fb.gaf.gz
+      entity_type:
+      status: active
+    species_code: Dmel
+    taxa:
+      - NCBITaxon:7227