Omics Integrator 2 Testing Error

[ntalluri]

No description provided.

[ntalluri]

There is now an error with Aggregate per algorithm for the ml step. It only runs when for algorithms with multiple parameter combinations chosen; however, some of those parameter combinations can lead to outputs that are still nothing, which can lead to the bug error from #117

[ntalluri]

The issue with the different headers in OI2 is nondeterministic. Although the code functions as intended, the corrupted data happens randomly, causing the ML option aggregate_per_algorithm to fail unpredictably if we keep the default as true.

[agitter]

Can you tell whether the non-determisitic behavior is the order of 'cost' and 'in_solution' or the presence of those columns? In #133 (comment) I speculated that perhpas the order of those two columns is allowed to be in either order, in which case our quick fix here could sort the columns as long as we get the correct number of columns.

[ntalluri]

I can check for the presence of the columns and order. I agree with the idea of sorting the columns as long as the correct number of columns and correct header names exist.

Every time I rerun SPRAS, it seems like the reordering happens at random when all the columns are present and is not specific to the dataset. With the current parameters in the config file, I haven't seen the missing in_solution column yet.

[ntalluri]

Ready for final review