Speed up list query

[cybersiddhu]

The list query seems to speed up quite a bit if the iteration is started from the cvterm instead of annotation collection. So, the existing one ...

FOR ann IN @@anno_collection
   FOR v IN 1..1 OUTBOUND ann GRAPH @anno_cvterm_graph
	FOR cv IN cv
		FILTER ann.is_obsolete == false
		FILTER v.deprecated == false
		FILTER v.label == @label
		FILTER v.graph_id == cv._id
		FILTER cv.metadata.namespace == @ontology
		LIMIT 10
		RETURN ann

should be replace by the following one

FOR cvt IN @@cvterm_collection
    FOR cv IN cv
        FOR ann IN 1..1 INBOUND cvt GRAPH "annotation_tag"
            FILTER cvt.graph_id == cv._id
            FILTER cvt.deprecated == false
            FILTER cvt.label == @label
            FILTER cv.metadata.namespace == @ontology
            FILTER ann.is_obsolete == false
            RETURN ann

[cybersiddhu]

• Rewrite the query
• Add an index on cvterm.label field.

[erichartline]

The updated query works quite well for most use cases. However, for the strain characteristics which does not include a label, it slows down drastically (5+ seconds per query).

To solve this, we discussed adding an identifier to the code (possibly called reverse) that would call the original query instead of the new one.