Use new support for prefixes in OBO format

Makefile

@@ -7,6 +7,8 @@ clean:
 
 TEST_SRCS ?= sgd pombase
 SRCS ?= sgd pombase mgi zfin rgd dictybase fb tair wb goa_human goa_human_complex goa_human_rna goa_human_isoform goa_pig xenbase pseudocap ecocyc goa_sars-cov-2 uniprot_reviewed
+ROBOT_ENV = ROBOT_JAVA_ARGS=-Xmx12G
+ROBOT = $(ROBOT_ENV) robot
 
 OBO_SRCS = $(patsubst %,target/neo-%.obo,$(SRCS))
 all_obo: $(OBO_SRCS)
@@ -23,8 +25,8 @@ trigger:
 	touch $@
 
 IMPORTS = imports/pr_import.obo
-neo.obo:  $(OBO_SRCS) $(IMPORTS)
-	owltools --create-ontology http://purl.obolibrary.org/obo/go/noctua/neo.owl $^ --merge-support-ontologies  -o -f obo [email protected] && grep -v ^owl-axioms [email protected] > $@
+neo.obo: $(OBO_SRCS) $(IMPORTS)
+	$(ROBOT) merge $(addprefix -i ,$^) annotate --ontology-iri 'http://purl.obolibrary.org/obo/go/noctua/neo.owl' convert -f obo -o [email protected] && grep -v ^owl-axioms [email protected] >$@
 
 ## datasets.json is created as a throwaway in the NEO versions of the
 ## pipeline and is based on the go-site master data.
@@ -75,20 +77,13 @@ target/neo-uniprot_reviewed.obo: mirror/uniprot_reviewed.gpi.gz
 # see below for regenerating this
 include Makefile-gafs
 
-# This is very hacky:
-#  - The neo solr index has an ID field (which is a CURIE), but no URI
-#  - Minerva requires OWL which uses URIs
+#  The neo solr index has an ID field (which is a CURIE), but no URI
+#  Minerva requires OWL which uses URIs
 #
 # When loading solr, owltools will use the oboInOwl:id field as priority to load the ID field (see https://github.com/owlcollab/owltools/pull/247)
 # Otherwise, the owltools built-in URI contraction method is used, which assumes OBO purls, with unpredictable behavior non-OBO PURLs
-#
-# Neo entities are NOT OBO ontologies, so they have a mix of prefixes, including identifiers.org
-#
-# Our hack is as follows. The perl code first generates an OBO file with CURIEs like FlyBase:FBgn111
-# The default owltools expansion makes this an OBO PURLs
-# We then "reverse" this with some hacky regexes...
 neo.owl: neo.obo
-	owltools $< -o [email protected] && ./bin/fix-obo-uris.pl [email protected] > [email protected] && mv $@.tmp2 $@
+	$(ROBOT) convert -i $< -o [email protected] -f owl && mv $@.tmp $@
 
 Makefile-gafs: datasets.json
 	./build-neo-makefile.py -i $< > [email protected] && mv [email protected] $@
@@ -110,7 +105,5 @@ rnacentral.gpi: rnacentral.gpi.gz
 target/neo-rnac.obo: rnacentral.gpi.gz
 	gzip -dc $< | ./rnacgpi2obo.pl > [email protected] && mv [email protected] $@
 
-target/xneo-%.owl: target/neo-%.obo
-	owltools $< -o [email protected] && mv [email protected] $@
-target/neo-%.owl: target/xneo-%.owl
-	./bin/fix-obo-uris.pl $< >  [email protected] && mv [email protected] $@
+target/neo-%.owl: target/neo-%.obo
+	$(ROBOT) convert -i $< -o [email protected] -f owl && mv [email protected] $@

gaf2obo.pl

@@ -5,6 +5,8 @@
 my $spn = 'generic';
 my $ontid;
 my $isoform_only = 0;
+open my $fh, '<', 'prefixes.obo.txt' or die "error opening prefixes.obo.txt: $!";
+my $prefixes = do { local $/; <$fh> };
 
 while (@ARGV) {
     my $opt = shift @ARGV;
@@ -24,6 +26,8 @@
 
 my $is_rat = $spn eq 'Rnor';
 
+print "format-version: 1.2\n";
+print $prefixes;
 print "ontology: go/noctua/$ontid\n";
 print "\n";
 

gpi2obo.pl

@@ -6,6 +6,8 @@
 my $fill_p = 0; # fill unknown species name with taxon id
 my $ontid;
 my $isoform_only = 0;
+open my $fh, '<', 'prefixes.obo.txt' or die "error opening prefixes.obo.txt: $!";
+my $prefixes = do { local $/; <$fh> };
 
 while (@ARGV) {
     my $opt = shift @ARGV;
@@ -26,6 +28,8 @@
     $ontid = $spn;
 }
 
+print "format-version: 1.2\n";
+print $prefixes;
 print "ontology: go/noctua/$ontid\n";
 print "\n";
 

prefixes.obo.txt

@@ -0,0 +1,15 @@
+idspace: AGI_LocusCode http://arabidopsis.org/servlets/TairObject?type=locus&name=
+idspace: ComplexPortal https://www.ebi.ac.uk/complexportal/complex/
+idspace: dictyBase http://identifiers.org/dictybase.gene/
+idspace: FB http://identifiers.org/flybase/
+idspace: GR_protein http://identifiers.org/uniprot/
+idspace: MGI http://identifiers.org/mgi/
+idspace: PomBase http://identifiers.org/pombase/
+idspace: RGD http://identifiers.org/rgd/
+idspace: RNAcentral http://rnacentral.org/rna/
+idspace: SGD http://identifiers.org/sgd/
+idspace: TAIR_locus http://identifiers.org/tair.locus/
+idspace: UniProtKB http://identifiers.org/uniprot/
+idspace: WB http://identifiers.org/wormbase/
+idspace: Xenbase http://identifiers.org/xenbase/
+idspace: ZFIN http://identifiers.org/zfin/

rnacgpi2obo.pl

@@ -18,8 +18,12 @@
 );
 
 my %tm = map {($_=>1)} @taxa;
-
 
+open my $fh, '<', 'prefixes.obo.txt' or die "error opening prefixes.obo.txt: $!";
+my $prefixes = do { local $/; <$fh> };
+
+print "format-version: 1.2\n";
+print $prefixes;
 print "ontology: go/noctua/rnac\n";
 print "\n";