TypeError: PLSRegression._single_perm() got an unexpected keyword argument 'samples'

[ChangJX521]

TypeError: PLSRegression._single_perm() got an unexpected keyword argument 'samples'

TO Repeoduce

from pyls import pls_regression
from netneurotools.stats import gen_spinsamples

n_spins=200

coords_dir = './parcellation/fsaverage/coords'

coords_lh = np.loadtxt(os.path.join(coords_dir, '1000Parcels_7Networks_coords_lh.txt'))
coords_rh = np.loadtxt(os.path.join(coords_dir, '1000Parcels_7Networks_coords_rh.txt'))

coords = np.vstack([coords_lh, coords_rh])
hemiid = np.array([0]*500 + [1]*500)


allspins = gen_spinsamples(coords, hemiid,
                                n_rotate=n_spins,
                                method='hungarian',
                                seed=42)


results = pls_regression(
    X, Y,
    n_components=1,
    n_perm=n_spins,
    n_boot=n_spins,
    permsamples=allspins,
    permindices=True,
    seed=42
)

Expected behavior

Permutation testing should run successfully, using the provided permsamples.

Actual behavior

Fails with a TypeError, because BasePLS.permutation() calls _single_perm(..., samples=...), but PLSRegression._single_perm() expects an argument named inds, not samples.

Proposed fix

Unify parameter names between BasePLS.permutation() and PLSRegression._single_perm().
Two possible solutions:

Change call in base.py from samples=... → inds=...

Or update PLSRegression._single_perm() signature to accept samples.

The below modification to PLSRegression._single_perm() would work

    def _single_perm(self, X, Y, samples, use_permind=True, groups=None, original=None, seed=None):
        """
        Permutes `Y` (w/o replacement) and recomputes PLS decomposition

        Parameters
        ----------
        X : (S, B) array_like
            Input data matrix, where `S` is observations and `B` is features
        Y : (S, T) array_like
            Input data matrix, where `S` is observations and `T` is features
        samples : (S,) array_like
            Resampling array
        groups : None
            Do nothing; here for compatibility purposes
        original : (B, N) array_like, optional
            Weights of `X` from original (non-permuted) decomposition; used to
            to align permuted weights
        seed : {int, :obj:`numpy.random.RandomState`, None}, optional
            Seed for random number generation. Default: None

        Returns
        -------
        varexp : (L,) `numpy.ndarray`
            Variance explained by PLS decomposition of permuted data
        """

        # should permute Y (but not X) by default
        Xp, Yp = self.make_permutation(X, Y, samples)
        x_weights, varexp, _ = self.svd(Xp, Yp, seed=seed)

        if self.inputs.rotate and original is not None:
            # flip signs of weights based on correlations with `original`
            flip = np.sign(compute.efficient_corr(x_weights, original))
            x_weights *= flip
            # NOTE: should we be doing a procrustes here?

            # recompute pctvar based on new x_weight signs
            mask = get_mask(Xp, Yp)
            y_loadings = Yp[mask].T @ (Xp @ x_weights)[mask]
            varexp = np.sum(y_loadings ** 2, axis=0) / np.sum(Yp[mask] ** 2)
        else:
            varexp = np.diag(varexp)

        # need to return len-3 for compatibility purposes
        return varexp, None, None