Advanced

Advanced#

Access the Vitessce configuration#

For advanced usage, you can obtain the VitessceConfig instance.

sc.pl.dotplot(adata, ...)

First, modify the plotting function call so that the VitessceWidget instance can be accessed via a variable:

vw = sc.pl.dotplot(adata, ...)
vw

The vw variable contains the VitessceWidget instance, which allows inspecting the VitessceConfig instance.

In a subsequent notebook cell:

# Access the configuration powering this widget instance.
vw.config
# or
vw.config.to_dict(base_url="")
# or, for "live" dynamically-updated config
vw._config

Access values from the coordination space#

For background, see the documentation on coordinated multiple views in Vitessce.

For instance, to access the list of currently-selected genes (i.e., features):

coordination_type = "featureSelection"
coordination_object = vw._config["coordinationSpace"][coordination_type]
coordination_object

Or, to access the set of currently-selected cell types (i.e., observation sets):

coordination_type = "obsSetSelection"
coordination_object = vw._config["coordinationSpace"][coordination_type]
coordination_object

See the full list of coordination types for more options.

When to use vitessce over easy_vitessce#

An advantage of using the plain vitessce Python package is that you do not need to have your AnnData/SpatialData object present/accessible. In fact, you can configure vitessce for hypothetical objects that do not exist or will never exist. This is useful if you want to construct configurations for objects that will be materialized later, e.g., in a data portal context. However, it also has the downside that you can easily construct an erroneous config (e.g., you specify a column name that does not exist in the targeted AnnData object) as the error will not arise until runtime/rendering-time.

On the other hand, easy_vitessce uses the object contents for certain logic. For example, when you specify color="GeneA" vs. color="cell_type" vs. color="red" in plotting params, we perform checks like:

if color in adata.var.index:
  # color was a gene name
  pass
elif color in adata.obs.columns:
  # color was an obs column name
  pass
else:
  # color may have been a color string or hex code string
  pass
Contents