Fix wrong sample list results from SQLAlchemy orm query

[yxu-lanl]

Closes microbiomedata/issues#1134

The problem is in Biosample query ORDER BY clause (order_by(desc(self.table.model.multiomics))). The 'multiomics' is not a unique column.

Solution:
Order by a unique column or a combination of columns that guarantees uniqueness.

Not sure ordering by name is better?
order_by(self.table.model.name, self.table.model.id)

[naglepuff]

Great catch! Thanks for taking this on.

I'm wondering if its valuable to preserve grouping biosamples by their data types (multiomics column). What if we sorted first by multiomics, and then something unique, like name or id? Maybe @aclum can weigh in on this.

[aclum]

I thought we were previously sorting on completed workflow results?

[naglepuff]

I thought we were previously sorting on completed workflow results?

I think that's what the multiomics column represents. Its a bitmap where each bit is a data type, so we set the corresponding bit to 1 if the biosample has workflow results for that data type. The precedence can be gleaned from looking at the bitmap definition.

[yxu-lanl]

The name is not unique. Will use id.
.order_by(self.table.model.multiomics.desc(), self.table.model.id) # type: ignore

nmdc_a=# select count(*) from biosample;
count

13248
(1 row)

nmdc_a=# select count(DISTINCT name) from biosample;
count

13049
(1 row)

nmdc_a=# select count(DISTINCT (multiomics, name)) from biosample;
count

13051
(1 row)

[yxu-lanl]

@naglepuff Could you approve this PR? Thanks.