control substitution on

statement ok
CREATE TABLE dna_sequences(sequence TEXT) AS VALUES
    ('ATCG'),
    ('AAA')
;

query I
SELECT reverse_complement(sequence) FROM dna_sequences
----
CGAT
TTT

query I
SELECT reverse_complement('ATCG')
----
CGAT

query I
SELECT reverse_complement('atcg')
----
cgat

statement ok
DROP TABLE dna_sequences;

query I
SELECT trim_polya('ATCGAAAA')
----
ATCG

query I
SELECT trim_polya('atcgaaaa')
----
atcg

query I
SELECT length(trim_polya('ATCGAAAA'))
----
4

query I
SELECT length(trim_polya('AAAA'))
----
0

statement ok
CREATE TABLE dna_sequences(s1 TEXT, s2 TEXT) AS VALUES ('ATCG', 'ATCG'), ('AAA', 'ATCG');

query I
SELECT length(trim_polya(s1)) FROM dna_sequences
----
4
0

query I
SELECT alignment_score('ATCG', 'ATCG')
----
4

query I
SELECT alignment_score('ATCG', 'AG')
----
1

query I
SELECT alignment_score('ATCG', 'ATCGG')
----
4

query I
SELECT alignment_score(s1, s2) score FROM dna_sequences
----
4
1

statement error
SELECT alignment_score('A', 'T', 'hi')

query I
SELECT locate_regex('agctggagctacc', 'agc')
----
[{start: 1, end: 4, match: agc}, {start: 7, end: 10, match: agc}]

query I
SELECT locate_regex('agctggagctacc', 'a[atcg]c')
----
[{start: 1, end: 4, match: agc}, {start: 7, end: 10, match: agc}, {start: 11, end: 14, match: acc}]

query I
SELECT locate_regex(s1, 'A[TCG]C') FROM dna_sequences
----
[{start: 1, end: 4, match: ATC}]
[]

statement error Execution error: locate_regex takes different types of arguments, the a sequence and a pattern
SELECT locate_regex('A[TCG]C', s1) FROM dna_sequences

statement ok
DROP TABLE dna_sequences;

query I
SELECT s1, integer_encoding(s1, 'ATCG') FROM (SELECT 'ATCG' s1 UNION ALL SELECT 'AAAA') t ORDER BY s1
----
AAAA [0, 0, 0, 0]
ATCG [0, 1, 2, 3]