test_image.py

import pytest

from anndata import AnnData
import scanpy as sc

import numpy as np

from matplotlib.testing.compare import compare_images
import matplotlib.pyplot as plt

from squidpy.im import ImageContainer
from tests.conftest import DPI, TOL, ACTUAL, EXPECTED, PlotTester, PlotTesterMeta
import squidpy as sq


class TestContainerShow(PlotTester, metaclass=PlotTesterMeta):
    def test_channelwise_wrong_number_of_axes(self, cont: ImageContainer):
        fig, ax = plt.subplots(dpi=DPI, tight_layout=True)
        with pytest.raises(ValueError, match=r"Expected `ax` to be of shape `\(1, 3\)`"):
            cont.show(ax=ax, channelwise=True)

    def test_plot_axis(self, cont: ImageContainer):
        cont.add_img(np.random.RandomState(42).normal(size=(*cont.shape, 3)), layer="foo")
        fig, (ax1, ax2) = plt.subplots(ncols=2, dpi=DPI, tight_layout=True)

        cont.show("image", ax=ax1)
        cont.show("foo", ax=ax2)

    def test_plot_channel(self, cont: ImageContainer):
        cont.show(channel=1, dpi=DPI)

    def test_plot_library_id(self, small_cont_4d: ImageContainer):
        small_cont_4d.show(library_id=["1"], dpi=DPI)

    def test_plot_segmentation(self, cont: ImageContainer):
        seg = np.random.RandomState(43).randint(0, 255, size=(*cont.shape, 1))
        seg[seg <= 200] = 0
        cont.add_img(seg, layer="foo")
        cont["foo"].attrs["segmentation"] = True

        cont.show("image", segmentation_layer="foo", dpi=DPI)

    def test_plot_imshow_kwargs(self, cont: ImageContainer):
        cont.show(channel=2, cmap="inferno", dpi=DPI)

    def test_plot_channelwise(self, cont: ImageContainer):
        cont.show(channelwise=True, dpi=DPI)

    def test_plot_channelwise_segmentation(self, cont: ImageContainer):
        seg = np.random.RandomState(43).randint(0, 255, size=(*cont.shape, 1))
        seg[seg <= 200] = 0
        cont.add_img(seg, layer="foo")
        cont["foo"].attrs["segmentation"] = True

        cont.show("image", channelwise=True, segmentation_layer="foo", dpi=DPI, segmentation_alpha=1)

    def test_plot_scale_mask_circle_crop(self, cont: ImageContainer):
        cont.crop_corner(0, 0, (200, 200), mask_circle=True, scale=2).show(dpi=DPI)

    @pytest.mark.parametrize("channelwise", [False, True])
    @pytest.mark.parametrize("transpose", [False, True])
    def test_transpose_channelwise(self, small_cont_4d: ImageContainer, transpose: bool, channelwise: bool):
        basename = f"{self.__class__.__name__[4:]}_transpose_channelwise_{transpose}_{channelwise}.png"
        small_cont_4d.show(transpose=transpose, channelwise=channelwise, dpi=DPI)

        plt.savefig(ACTUAL / basename, dpi=DPI)
        plt.close()
        res = compare_images(str(EXPECTED / basename), ACTUAL / basename, TOL)

        assert res is None, res


def test_extract(adata: AnnData, cont: ImageContainer, caplog):
    """
    Calculate features and extract columns to obs
    """
    # get obsm
    sq.im.calculate_image_features(adata, cont, features=["summary"])

    # extract columns (default values)
    extr_adata = sq.pl.extract(adata)
    # Test that expected columns exist
    for col in [
        "summary_ch-0_quantile-0.9",
        "summary_ch-0_quantile-0.5",
        "summary_ch-0_quantile-0.1",
        "summary_ch-1_quantile-0.9",
        "summary_ch-1_quantile-0.5",
        "summary_ch-1_quantile-0.1",
        "summary_ch-2_quantile-0.9",
        "summary_ch-2_quantile-0.5",
        "summary_ch-2_quantile-0.1",
    ]:
        assert col in extr_adata.obs.columns

    # get obsm that is a numpy array
    adata.obsm["pca_features"] = sc.pp.pca(np.asarray(adata.obsm["img_features"]), n_comps=3)
    # extract columns
    extr_adata = sq.pl.extract(adata, obsm_key="pca_features", prefix="pca_features")
    # Test that expected columns exist
    for col in ["pca_features_0", "pca_features_1", "pca_features_2"]:
        assert col in extr_adata.obs.columns

    # extract multiple obsm at once (no prefix)
    extr_adata = sq.pl.extract(adata, obsm_key=["img_features", "pca_features"])
    # Test that expected columns exist
    for col in [
        "summary_ch-0_quantile-0.9",
        "summary_ch-0_quantile-0.5",
        "summary_ch-0_quantile-0.1",
        "summary_ch-1_quantile-0.9",
        "summary_ch-1_quantile-0.5",
        "summary_ch-1_quantile-0.1",
        "summary_ch-2_quantile-0.9",
        "summary_ch-2_quantile-0.5",
        "summary_ch-2_quantile-0.1",
        "0",
        "1",
        "2",
    ]:
        assert col in extr_adata.obs.columns

    # TODO: test similarly to ligrec
    # currently logging to stderr, and not captured by caplog
    # extract obsm twice and make sure that warnings are issued
    # with caplog.at_level(logging.WARNING):
    #    extr2_adata = sq.pl.extract(extr_adata, obsm_key=['pca_features'])
    #    log = caplog.text
    #    assert "will be overwritten by extract" in log