Technology focus: 10x Genomics Visium

This notebook will present a rough overview of the plotting functionalities that spatialdata implements for Visium data.

Loading the data

Please download the data from here: Visium dataset and rename it (eventually using symlinks) to visium_brain.zarr.

visium_zarr_path = "./visium_brain.zarr"

import spatialdata as sd

visium_sdata = sd.read_zarr(visium_zarr_path)
visium_sdata

SpatialData object with:
├── Images
│     ├── 'ST8059048_hires_image': SpatialImage[cyx] (3, 2000, 1969)
│     ├── 'ST8059048_lowres_image': SpatialImage[cyx] (3, 600, 591)
│     ├── 'ST8059050_hires_image': SpatialImage[cyx] (3, 2000, 1968)
│     └── 'ST8059050_lowres_image': SpatialImage[cyx] (3, 600, 590)
├── Shapes
│     ├── 'ST8059048': GeoDataFrame shape: (4992, 2) (2D shapes)
│     └── 'ST8059050': GeoDataFrame shape: (3497, 2) (2D shapes)
└── Tables
      └── 'table': AnnData (8489, 31053)
with coordinate systems:
▸ 'ST8059048', with elements:
        ST8059048_hires_image (Images), ST8059048 (Shapes)
▸ 'ST8059050', with elements:
        ST8059050_hires_image (Images), ST8059050 (Shapes)

Visualise the data

We’re going to create a naiive visualisation of the data, overlaying the Visium spots and the tissue images. For this, we need to load the spatialdata_plot library which extends the sd.SpatialData object with the .pl module.

import spatialdata_plot

visium_sdata.pl.render_images().pl.render_shapes().pl.show("ST8059050")

We can see that the data contains two coordinate systems (ST8059050 and ST8059052) with image and spot information each. In SpatialData, these spots are represented as Shapes. When giving no further parameters, one panel is generated per coordinate system with the members that have been specified in the function call. We can see that the spots are aligned to the tissue representation which is also respected by the plotting logic.

However, the spots are all grey since we have not provided any information on what they should encode. Such information can be found in the Table attribute (which is an anndata.AnnData table) of the SpatialData object, either in the data itself or the obs attribute.

visium_sdata["table"].to_df().sum(axis=0).sort_values(ascending=False).head(10)
# We will select some of the highly expressed genes for this example

mt-Co3     3366555.0
mt-Co1     3134006.0
mt-Atp6    2175621.0
mt-Co2     2160662.0
mt-Cytb    1317073.0
mt-Nd4     1100399.0
mt-Nd1     1098630.0
Fth1        840543.0
Ttr         833197.0
mt-Nd2      775515.0
dtype: float32

visium_sdata["table"].obs.head(3)

	in_tissue	array_row	array_col	spot_id	region
AAACAACGAATAGTTC-1	0	0	16	0	ST8059048
AAACAAGTATCTCCCA-1	1	50	102	1	ST8059048
AAACAATCTACTAGCA-1	0	3	43	2	ST8059048

Color the visium spots by gene expression

To use this information in our plot, we pass the name of the column by which we want to color our expression to color. Furthermore, we are going to subset the data to only one coordinate system.

(
    visium_sdata.pp.get_elements("ST8059050")
    .pl.render_images(elements="ST8059050_hires_image")
    .pl.render_shapes(elements="ST8059050", color="mt-Co3")
    .pl.show()
)

We can also provide ax objects to spatialdata_plot for further customisation.

import matplotlib.pyplot as plt

fig, axs = plt.subplots(ncols=3, nrows=1, figsize=(12, 3))

visium_sdata_subset = visium_sdata.pp.get_elements("ST8059050")

visium_sdata_subset.pl.render_shapes(elements="ST8059050", color="mt-Co1").pl.show(ax=axs[0], title="mt-Co1")

visium_sdata_subset.pl.render_shapes(elements="ST8059050", color="Fth1").pl.show(ax=axs[1], title="Fth1")

visium_sdata_subset.pl.render_shapes(elements="ST8059050", color="Ttr").pl.show(ax=axs[2], title="Ttr")

plt.tight_layout()