What are you trying to achieve?

If the goal is to load/view data locally for development you could use InteractiveRunner which allows materializing data as a pandas dataframe https://cloud.google.com/dataflow/docs/guides/interactive-pipeline-development

beam.combiners.ToList collects all elements in a PCollection into a single list element in the output PCollection [1]. If the goal is to process a PCollection as a single list element this is suitable.

If your goal is to load the output of the dataset as a list in your local program after pipeline execution it requires writing to some kind of storage and then reading in your local program.

[1] https://beam.apache.org/documentation/transforms/python/aggregation/tolist/