Est speedup

Model/lib/psql/webready/orgSpecific/EstAlignmentGeneSummary.psql

@@ -18,7 +18,8 @@
              dots.NaSequence query_sequence 
         WHERE e.na_sequence_id = ba.query_na_sequence_id
           AND ga.na_sequence_id = ba.target_na_sequence_id
-          AND least(ba.target_end, ga.gene_end_max) - greatest(ba.target_start, ga.gene_start_min) >= 0
+          AND ba.target_end >= ga.gene_start_min
+          AND ga.gene_end_max >= ba.target_start
           AND query_sequence.na_sequence_id = ba.query_na_sequence_id
           AND ga.org_abbrev = ':ORG_ABBREV'
 
@@ -50,11 +51,15 @@
             /* (because they overlap no genes) */
             SELECT ba.blat_alignment_id
             FROM dots.BlatAlignment ba, dots.NaSequence query_sequence,
-                 sres.OntologyTerm so, dots.NaSequence target_sequence
+                 sres.OntologyTerm so, dots.NaSequence target_sequence,
+                :SCHEMA.taxonspecies tsq, :SCHEMA.taxonspecies tst
             WHERE query_sequence.na_sequence_id = ba.query_na_sequence_id
               AND query_sequence.sequence_ontology_id = so.ontology_term_id
               AND ba.target_na_sequence_id = target_sequence.na_sequence_id
-              AND target_sequence.taxon_id = :TAXON_ID
+              /* we filter the query sequence (at the species level) because doing so provides a dramatic performance boost */
+              AND query_sequence.taxon_id = tsq.taxon_id
+              AND tst.species_taxon_id = tsq.species_taxon_id
+              AND tst.taxon_id = :TAXON_ID
               AND so.name = 'EST'
             EXCEPT
             SELECT blat_alignment_id FROM :SCHEMA.:CLEAN_ORG_ABBREVEstAlignmentGeneTmp)

Model/lib/psql/webready/orgSpecific/EstAttributes.psql

@@ -21,28 +21,13 @@
                       END, 1, 100) AS organism,
                taxon.ncbi_tax_id,
                ed.name AS external_db_name,
-               coalesce(best.best_alignment_count, 0) AS best_alignment_count,
                l.library_id, replace(l.dbest_name, '''', '-') as library_dbest_name
         FROM dots.Est e, dots.Library l, sres.Taxon, sres.OntologyTerm oterm,
              sres.TaxonName tn, sres.ExternalDatabase ed,
              apidb.datasource ds, apidb.organism o,
+             :SCHEMA.taxonspecies ts_r,  -- ref organism
+             :SCHEMA.taxonspecies ts_e,  -- est organism
              sres.ExternalDatabaseRelease edr, dots.ExternalNaSequence ens
-             LEFT JOIN
-             (select query_na_sequence_id,max(ct) as best_alignment_count
-              from (
-                    SELECT query_na_sequence_id, COUNT(*) AS ct
-                     FROM dots.BlatAlignment ba, apidb.datasource ds, apidb.organism o,
-                     sres.externaldatabase d, sres.externaldatabaserelease r
-                     WHERE is_best_alignment = 1
-                     AND ba.query_external_db_release_id = r.external_database_release_id
-                     AND r.external_database_id = d.external_database_id
-                     AND d.name = ds.name
-                     AND ds.taxon_id = o.taxon_id
-                     AND o.is_reference_strain = 1
-                     AND o.taxon_id = :TAXON_ID
-                     GROUP BY target_external_db_release_id,query_na_sequence_id) t
-              group by query_na_sequence_id
-              ) best ON ens.na_sequence_id = best.query_na_sequence_id
         WHERE e.na_sequence_id = ens.na_sequence_id
           AND e.library_id = l.library_id
           AND ens.taxon_id = tn.taxon_id
@@ -52,8 +37,11 @@
           AND edr.external_database_id = ed.external_database_id
           AND ens.sequence_ontology_id = oterm.ontology_term_id
           AND ed.name = ds.name
+          AND ens.taxon_id = ts_e.taxon_id
+          AND o.taxon_id = ts_r.taxon_id
+          AND ts_e.species_taxon_id = ts_r.species_taxon_id
           and ds.taxon_id = o.taxon_id
-          and o.is_reference_strain = 1
+          and o.is_reference_strain = 1 -- We want ESTs for the species of the reference strain 
           and o.taxon_id = :TAXON_ID
           AND oterm.name = 'EST';
 :DECLARE_PARTITION;