SQLAlchemy filter in_ operator

Question

I am trying to do a simple filter operation on a query in sqlalchemy, like this:

q = session.query(Genotypes).filter(Genotypes.rsid.in_(inall))

Answer 1

The below workaround worked for me:

q = session.query(Genotypes).filter(Genotypes.rsid.in_(inall))
query_as_string = str(q.statement.compile(compile_kwargs={"literal_binds": True}))
session.execute(query_as_string).first()

This basically forces the query to compile as a string before execution, which bypasses the whole variables issue. Some details on this are available in SQLAlchemy's docs here.

BTW, if you're not using SQLite you can make use of the ANY operator to pass the list object as a single parameter (see my answer to this question here).

Answer 2

If the table where you are getting your rsids from is available in the same database I'd use a subquery to pass them into your Genotypes query rather than passing the one million entries around in your Python code.

sq = session.query(RSID_Source).subquery()
q = session.query(Genotypes).filter(Genotypes.rsid.in_(sq))

The issue is that in order to pass that list to SQLite (or any database, really), SQLAlchemy has to pass over each entry for your in clause as a variable. The SQL translates roughly to:

-- Not valid SQLite SQL
DECLARE @Param1 TEXT;
SET @Param1 = ?;
DECLARE @Param2 TEXT;
SET @Param2 = ?;
-- snip 999,998 more

SELECT field1, field2, -- etc.
FROM Genotypes G
WHERE G.rsid IN (@Param1, @Param2, /* snip */)