Parse H5 backed delayed objects (#50)

* Update tests, documentation and README

README.md

@@ -43,15 +43,9 @@ Package is published to [PyPI](https://pypi.org/project/rds2py/)
 
 ```shell
 pip install rds2py
-```
-
-## Quick Start
-
-```python
-from rds2py import read_rds
 
-# Read any RDS file
-r_obj = read_rds("path/to/file.rds")
+# or install optional dependencies
+pip install rds2py[optional]
 ```
 
 ## Usage

setup.cfg

@@ -51,11 +51,11 @@ install_requires =
     importlib-metadata; python_version<"3.8"
     numpy
     scipy
+    biocframe
     biocutils>=0.1.5
-    singlecellexperiment>=0.4.1
-    summarizedexperiment>=0.4.1
     genomicranges>=0.4.9
-    biocframe
+    summarizedexperiment>=0.4.1
+    singlecellexperiment>=0.4.1
     multiassayexperiment
 
 [options.packages.find]
@@ -69,12 +69,15 @@ exclude =
 # PDF = ReportLab; RXP
 optional =
     pandas
+    hdf5array
 
 # Add here test requirements (semicolon/line-separated)
 testing =
     setuptools
     pytest
     pytest-cov
+    pandas
+    hdf5array
 
 [options.entry_points]
 # Add here console scripts like:

src/rds2py/generics.py

@@ -58,6 +58,8 @@
     # multi assay experiment
     "MultiAssayExperiment": "rds2py.read_mae.read_multi_assay_experiment",
     "ExperimentList": "rds2py.read_dict.read_dict",
+    # delayed matrices
+    "H5SparseMatrix": "rds2py.read_delayed_matrix.read_hdf5_sparse",
 }
 
 

src/rds2py/read_delayed_matrix.py

@@ -0,0 +1,41 @@
+"""Functions and classes for parsing R delayed matrix objects from HDF5Array."""
+
+from hdf5array import Hdf5CompressedSparseMatrix
+
+from .generics import _dispatcher
+from .rdsutils import get_class
+
+__author__ = "jkanche"
+__copyright__ = "jkanche"
+__license__ = "MIT"
+
+
+def read_hdf5_sparse(robject: dict, **kwargs) -> Hdf5CompressedSparseMatrix:
+    """Convert an R delayed sparse array (H5-backed).
+
+    Args:
+        robject:
+            Dictionary containing parsed delayed sparse array.
+
+        **kwargs:
+            Additional arguments.
+
+    Returns:
+       A Hdf5CompressedSparseMatrix from the 'hdf5array' package.
+    """
+    _cls = get_class(robject)
+    if _cls not in ["H5SparseMatrix"]:
+        raise RuntimeError(f"`robject` does not contain not a 'H5SparseMatrix' object, contains `{_cls}`.")
+
+    by_column = False
+    # get seed package name
+    _seed_cls = get_class(robject["attributes"]["seed"])
+    if _seed_cls in ["CSC_H5SparseMatrixSeed"]:
+        by_column = True
+
+    _seed_obj = robject["attributes"]["seed"]
+    shape = tuple(_dispatcher(_seed_obj["attributes"]["dim"], **kwargs))
+    fpath = list(_dispatcher(_seed_obj["attributes"]["filepath"], **kwargs))[0]
+    group_name = list(_dispatcher(_seed_obj["attributes"]["group"], **kwargs))[0]
+
+    return Hdf5CompressedSparseMatrix(path=fpath, group_name=group_name, shape=shape, by_column=by_column)

tests/data/generate_files.R

@@ -177,3 +177,13 @@ simpleMultiAssay <- MultiAssayExperiment(experiments=doubleExp)
 simpleMultiAssay2 <- MultiAssayExperiment(experiments=doubleExp,
                                           colData=patient.data)
 saveRDS(simpleMultiAssay2, "simple_mae.rds")
+
+## Delayed Arrays
+
+library(zellkonverter)
+h5ad_file <- system.file("extdata", "example_anndata.h5ad",
+                         package="zellkonverter")
+h5ls(h5ad_file)
+
+M <- H5SparseMatrix(h5ad_file, "/obsp/connectivities")
+saveRDS(M, "h5sparse.rds")

tests/test_delayedmatrices.py

@@ -0,0 +1,16 @@
+import pytest
+
+from rds2py import read_rds
+from hdf5array import Hdf5CompressedSparseMatrix
+
+__author__ = "jkanche"
+__copyright__ = "jkanche"
+__license__ = "MIT"
+
+@pytest.mark.skip(reason="delayedarray uses full file paths. this should be run locally.")
+def test_read_h5sparse():
+    array = read_rds("tests/data/h5sparse.rds")
+
+    assert array is not None
+    assert isinstance(array, Hdf5CompressedSparseMatrix)
+    assert array.shape == (200, 200)