UnboundLocalError when using design matrix

[dawe]

I'm trying to use a design matrix to run a simple test, in place of a formula but I cannot use due to this error:

>>> dmat_flat = de.utils.design_matrix(data = adata, formula="~0 + groups")
>>> test_flat = de.test.wald(
data = adata,
dmat_loc=dmat_flat[0],
)

UnboundLocalError                         Traceback (most recent call last)
<ipython-input-167-850899a8ac49> in <module>
      1 test_flat = de.test.wald(
      2 data = adata,
----> 3 dmat_loc=dmat_flat[0],
      4 
      5 )

~/anaconda3/envs/graph-tool/lib/python3.7/site-packages/diffxpy/testing/tests.py in wald(data, factor_loc_totest, coef_to_test, formula_loc, formula_scale, as_numeric, init_a, init_b, gene_names, sample_description, dmat_loc, dmat_scale, constraints_loc, constraints_scale, noise_model, size_factors, batch_size, backend, train_args, training_strategy, quick_scale, dtype, **kwargs)
    651         as_numeric=as_numeric,
    652         constraints=constraints_loc,
--> 653         return_type="patsy"
    654     )
    655     design_scale, design_scale_names, constraints_scale, term_names_scale = constraint_system_from_star(

~/anaconda3/envs/graph-tool/lib/python3.7/site-packages/diffxpy/testing/utils.py in constraint_system_from_star(dmat, sample_description, formula, as_numeric, constraints, return_type)
    270         as_categorical=as_categorical,
    271         constraints=constraints,
--> 272         return_type=return_type
    273     )
    274 

~/anaconda3/envs/graph-tool/lib/python3.7/site-packages/batchglm/data.py in constraint_system_from_star(dmat, sample_description, formula, as_categorical, constraints, return_type)
    259             )
    260 
--> 261     return dmat, coef_names, cmat, term_names
    262 
    263 

UnboundLocalError: local variable 'coef_names' referenced before assignment

Note that I'm not sure how to pass a design matrix in place of a formula, I was performing some usage test...

[davidsebfischer]

You still have to declare a coefficient that you want to test right now you just gave .wald() instructions for fitting a GLM but not which coefficient to test. I guess you want to test the group parameter, you have to define this as a subset of all model parameters using factor_loc_totest or coef_to_test!

[dawe]

I have the very same error if coef_to_test is specified (btw, factor_loc_totest is not supported when dmat_loc is given)

[davidsebfischer]

I have the very same error if coef_to_test is specified (btw, factor_loc_totest is not supported when dmat_loc is given)

can you post the exact code you are using when specifying coef_to_test? for now could just give wald the formula instead of building it yourself first? that makes parsing easier, also the reason via factors are not supported in this mode

[dawe]

Given the design matrix above

dmat_flat = de.utils.design_matrix(data = adata, formula="~0 + groups")

I use the list of coefficients names in the second element of the tuple

de.test.wald(
    data = adata,
    dmat_loc=dmat_flat[0], 
    coef_to_test=dmat_flat[1])

or any subset of it