Save and load MultiAssayExperiments in Python

Introduction

The dolomite-mae package is the Python counterpart to the alabaster.mae R package, providing methods for saving/reading MultiAssayExperiment objects within the dolomite framework. All components of the MultiAssayExperiment - column_data, sample map and experiments - are saved to their respective file representations, which can be loaded in a new R/Python environment for cross-language analyses.

Quick start

Let's mock up a MultiAssayExperiment:

from multiassayexperiment import MultiAssayExperiment
from singlecellexperiment import SingleCellExperiment
from summarizedexperiment import SummarizedExperiment
import biocframe
import numpy

x = numpy.random.rand(1000, 200)
x2 = (numpy.random.rand(1000, 200) * 10).astype(numpy.int32)

sce = SingleCellExperiment(
     {"logcounts": x, "counts": x2},
     main_experiment_name="aaron's secret modality",
     row_data=biocframe.BiocFrame(
          {"foo": numpy.random.rand(1000), "bar": numpy.random.rand(1000)},
          row_names=["gene_sce_" + str(i) for i in range(1000)],
     ),
     column_data=biocframe.BiocFrame(
          {"whee": numpy.random.rand(200), "stuff": numpy.random.rand(200)},
          row_names=["cell_sce" + str(i) for i in range(200)],
     ),
)

se = SummarizedExperiment(
     {"counts": numpy.random.rand(100, 200)},
     row_data=biocframe.BiocFrame(
          {"foo": numpy.random.rand(100), "bar": numpy.random.rand(100)},
          row_names=["gene_se_" + str(i) for i in range(100)],
     ),
     column_data=biocframe.BiocFrame(
          {"whee": numpy.random.rand(200), "stuff": numpy.random.rand(200)},
          row_names=["cell_se" + str(i) for i in range(200)],
     ),
)

mae = MultiAssayExperiment(experiments={"jay_expt": sce, "aarons_expt": se})

Now we can save it:

from dolomite_base import save_object
import dolomite_se
import os
from tempfile import mkdtemp

path = os.path.join(mkdtemp(), "test")
save_object(se, path)

And load it again, e,g., in a new session:

from dolomite_base import read_object

roundtrip = read_object(path)