| 1 |
5491
|
aaronmk
|
SELECT *
|
| 2 |
5721
|
aaronmk
|
FROM "nodes.src"
|
| 3 |
|
|
JOIN "names.src" USING (tax_id)
|
| 4 |
5491
|
aaronmk
|
WHERE "name class" = 'scientific name'
|
| 5 |
|
|
;
|
| 6 |
|
|
|
| 7 |
|
|
ALTER TABLE :table ALTER COLUMN tax_id TYPE integer
|
| 8 |
|
|
USING tax_id::integer;
|
| 9 |
|
|
ALTER TABLE :table ALTER COLUMN "parent tax_id" TYPE integer
|
| 10 |
|
|
USING "parent tax_id"::integer;
|
| 11 |
|
|
|
| 12 |
|
|
ALTER TABLE :table ADD PRIMARY KEY (tax_id);
|
| 13 |
5820
|
aaronmk
|
|
| 14 |
|
|
CREATE INDEX ON :table ("parent tax_id");
|
| 15 |
|
|
|
| 16 |
|
|
ALTER TABLE :table ADD FOREIGN KEY ("parent tax_id") REFERENCES :table (tax_id)
|
| 17 |
|
|
ON UPDATE CASCADE ON DELETE CASCADE;
|
| 18 |
5821
|
aaronmk
|
|
| 19 |
6033
|
aaronmk
|
-- Make name_txt (mostly) globally unique by removing other kingdoms
|
| 20 |
6032
|
aaronmk
|
-- Note that the delete will cascade to descendants
|
| 21 |
6033
|
aaronmk
|
DELETE FROM :table WHERE tax_id IN (
|
| 22 |
|
|
10239 -- superkingdom Viruses
|
| 23 |
|
|
, 12884 -- superkingdom Viroids
|
| 24 |
|
|
, 12908 -- unclassified sequences
|
| 25 |
|
|
, 28384 -- other sequences
|
| 26 |
|
|
, 2 -- superkingdom Bacteria
|
| 27 |
|
|
, 2157 -- superkingdom Archaea
|
| 28 |
|
|
, 4751 -- kingdom Fungi
|
| 29 |
|
|
, 33208 -- kingdom Animalia/Metazoa
|
| 30 |
|
|
);
|
| 31 |
5835
|
aaronmk
|
|
| 32 |
6034
|
aaronmk
|
-- Make name_txt completely globally unique by removing duplicates
|
| 33 |
5835
|
aaronmk
|
DELETE FROM :table
|
| 34 |
6034
|
aaronmk
|
WHERE name_txt IN (
|
| 35 |
5835
|
aaronmk
|
SELECT name_txt
|
| 36 |
|
|
FROM :table
|
| 37 |
|
|
GROUP BY name_txt
|
| 38 |
|
|
HAVING count(*) > 1
|
| 39 |
|
|
)
|
| 40 |
|
|
;
|
