Implement get/set for row/column data and coercions betwee SE/RSE (#103)

Author: jkanche

CHANGELOG.md

@@ -1,10 +1,12 @@
 # Changelog
 
-## Version 0.6.0 - 0.6.2
+## Version 0.6.0 - 0.6.3
 
 - Classes now extend `BiocObject` from biocutils, which provides a metadata field.
 - `valdiate` is renamed to `_validate` for consistency with other classes and packages.
 - Update github actions to run from 3.10 - 3.14.
+- Coerce: RSE to/from SE and vice-versa.
+- Implement getters and setters to extract columns from either row/column data (See [issue](https://github.com/BiocPy/SummarizedExperiment/issues/99)).
 
 ## Version 0.5.4 - 0.5.5
 

src/summarizedexperiment/RangedSummarizedExperiment.py

@@ -942,18 +942,55 @@ def combine_columns(self, *other) -> RangedSummarizedExperiment:
     ######>> to se <<#####
     ######################
 
+    @classmethod
+    def from_summarizedexperiment(cls, se: SummarizedExperiment) -> RangedSummarizedExperiment:
+        return cls(
+            assays=se.get_assays(),
+            row_data=se.get_row_data(),
+            column_data=se.get_column_data(),
+            row_names=se.get_row_names(),
+            column_names=se.get_column_names(),
+            metadata=se.get_metadata(),
+        )
+
+    @classmethod
+    def from_se(cls, se: SummarizedExperiment) -> RangedSummarizedExperiment:
+        """Alias for :py:meth:`~to_summarizedexperiment`."""
+        return RangedSummarizedExperiment.from_summarizedexperiment(se)
+
     def to_summarizedexperiment(self):
         """Coerce to :py:class:`summarizedexperiment.SummarizedExperiment.SummarizedExperiment`"""
 
-        warn("SE does not containg a slot for ranges, dropping ranges.", UserWarning)
+        warn("SE does not containg a slot for ranges, merging ranges with row_data.", UserWarning)
+
+        new_row_data = self._rows
+        if self._row_ranges is not None:
+            try:
+                import copy
+
+                if isinstance(self._row_ranges, CompressedGenomicRangesList):
+                    df = self._row_ranges.get_unlist_data().to_pandas()
+                else:
+                    df = self._row_ranges.to_pandas()
+
+                range_bf = biocframe.BiocFrame.from_pandas(df)
+                if new_row_data is None or new_row_data.shape[0] == 0:
+                    new_row_data = range_bf
+                else:
+                    new_row_data = copy.copy(new_row_data)
+                    for col in range_bf.column_names:
+                        new_row_data[col] = range_bf[col]
+            except Exception as e:
+                warn(f"Failed to transfer row_ranges to row_data: {e}", UserWarning)
 
         return SummarizedExperiment(
             assays=self._assays,
-            row_data=self._rows,
+            row_data=new_row_data,
             column_data=self._cols,
             row_names=self._row_names,
             column_names=self._column_names,
             metadata=self._metadata,
+            _validate=False,
         )
 
     def to_se(self):

src/summarizedexperiment/SummarizedExperiment.py

@@ -128,13 +128,33 @@ def combine_columns(self, *other) -> SummarizedExperiment:
     ######>> to rse <<#####
     #######################
 
-    def to_rangedsummarizedexperiment(self):
-        """Coerce to :py:class:`summarizedexperiment.RangedSummarizedExperiment.RangedSummarizedExperiment`"""
+    def to_rangedsummarizedexperiment(self, row_ranges=None):
+        """Coerce to :py:class:`summarizedexperiment.RangedSummarizedExperiment.RangedSummarizedExperiment`.
+
+        Args:
+            row_ranges:
+                Genomic features, must be the same length as the number of rows of
+                the matrices in assays.
+
+                If ``row_ranges`` is not provided, this method will attempt to
+                extract range information (e.g., 'seqnames', 'starts') from
+                ``row_data`` to construct the :py:class:`~genomicranges.GenomicRanges.GenomicRanges`.
+        """
 
         from .RangedSummarizedExperiment import RangedSummarizedExperiment
 
+        if row_ranges is None:
+            try:
+                rd_cols = self.row_data.get_column_names()
+                if "seqnames" in rd_cols and "starts" in rd_cols:
+                    df = self.row_data.to_pandas()
+                    row_ranges = GenomicRanges.from_pandas(df)
+            except Exception as _:
+                pass
+
         return RangedSummarizedExperiment(
             assays=self._assays,
+            row_ranges=row_ranges,
             row_data=self._rows,
             column_data=self._cols,
             row_names=self._row_names,
@@ -143,9 +163,9 @@ def to_rangedsummarizedexperiment(self):
             _validate=False,
         )
 
-    def to_rse(self):
+    def to_rse(self, row_ranges=None):
         """Alias for :py:meth:`~to_rangedsummarizedexperiment`."""
-        return self.to_rangedsummarizedexperiment()
+        return self.to_rangedsummarizedexperiment(row_ranges=row_ranges)
 
 
 ############################

src/summarizedexperiment/base.py

@@ -1197,3 +1197,40 @@ def to_anndata(self):
         )
 
         return obj
+
+    ######################################
+    ######>> col_data accessors <<########
+    ######################################
+
+    def get_column_data_column(self, column: str) -> Any:
+        """Access a column from the ``column_data``.
+
+        Args:
+            column:
+                Name of the column to retrieve.
+
+        Returns:
+            The content of the column.
+        """
+        return self._cols.get_column(column)
+
+    def set_column_data_column(self, column: str, value: Any, in_place: bool = False) -> BaseSE:
+        """Set or replace a column in ``column_data``.
+
+        Args:
+            column:
+                Name of the column to set.
+
+            value:
+                Values for the column. Must match the number of samples (columns).
+
+            in_place:
+                Whether to modify the ``BaseSE`` in place.
+
+        Returns:
+            A modified ``BaseSE`` object, either as a copy of the original
+            or as a reference to the (in-place-modified) original.
+        """
+        output = self._define_output(in_place)
+        output._cols = output._cols.set_column(column, value, in_place=False)
+        return output

tests/test_RSE.py

@@ -5,6 +5,7 @@
 import numpy as np
 import pandas as pd
 import pytest
+from summarizedexperiment.SummarizedExperiment import SummarizedExperiment
 from summarizedexperiment.RangedSummarizedExperiment import RangedSummarizedExperiment
 
 __author__ = "jkanche"
@@ -132,3 +133,16 @@ def test_RSE_empty():
 
     assert tse.row_names is None
     assert tse.col_names is None
+
+def test_RSE_from_SE():
+    tse = SummarizedExperiment(
+        assays={"counts": counts}, row_data=df_gr, column_data=col_data
+    )
+
+    assert tse is not None
+    assert isinstance(tse, SummarizedExperiment)
+    assert tse.shape == (200, 6)
+
+    rse = RangedSummarizedExperiment.from_se(tse)
+    assert isinstance(rse, RangedSummarizedExperiment)
+    assert rse.shape == (200, 6)

tests/test_RSE_methods.py

@@ -231,3 +231,19 @@ def test_RSE_to_se():
     assert se is not None
     assert isinstance(se, SummarizedExperiment)
     assert rse.shape == se.shape
+
+def test_RSE_to_se_with_ranges():
+    tse = RangedSummarizedExperiment(
+        assays={"counts": counts}, row_ranges=gr, column_data=col_data
+    )
+
+    se = tse.to_summarizedexperiment()
+
+    assert se is not None
+    assert isinstance(se, SummarizedExperiment)
+    assert se.shape == tse.shape
+
+    assert se.row_data is not None
+    assert "seqnames" in se.row_data.column_names
+    assert "starts" in se.row_data.column_names
+    assert "ends" in se.row_data.column_names

tests/test_SE.py

@@ -4,6 +4,7 @@
 import numpy as np
 import pandas as pd
 from summarizedexperiment.SummarizedExperiment import SummarizedExperiment
+from summarizedexperiment.RangedSummarizedExperiment import RangedSummarizedExperiment
 import delayedarray
 from anndata import AnnData
 

tests/test_SE_methods.py

@@ -149,3 +149,41 @@ def test_SE_to_rse():
     assert rse is not None
     assert isinstance(rse, RangedSummarizedExperiment)
     assert rse.shape == tse.shape
+
+def test_SE_to_rse_parse_ranges():
+    rd = BiocFrame({
+        "seqnames": ["chr1"] * nrows,
+        "starts": range(nrows),
+        "ends": range(nrows),
+        "strand": ["+"] * nrows
+    })
+
+    tse = SummarizedExperiment(
+        assays={"counts": counts}, row_data=rd, column_data=col_data
+    )
+
+    rse = tse.to_rangedsummarizedexperiment()
+
+    assert rse is not None
+    assert isinstance(rse, RangedSummarizedExperiment)
+    assert rse.shape == tse.shape
+
+    assert rse.row_ranges is not None
+    assert len(rse.row_ranges) == nrows
+    assert np.allclose(rse.row_ranges.start, np.array(range(nrows)))
+
+def test_SE_coldata_accessors():
+    tse = SummarizedExperiment(
+        assays={"counts": counts}, row_data=row_data, column_data=col_data
+    )
+
+    assert tse.get_column_data_column("treatment") is not None
+    assert len(tse.get_column_data_column("treatment")) == 6
+
+    new_tse = tse.set_column_data_column("stuff", [1, 2, 3, 4, 5, 6])
+    assert new_tse.shape == tse.shape
+    assert "stuff" in new_tse.col_data.column_names
+    assert "stuff" not in tse.col_data.column_names
+
+    tse.set_column_data_column("stuff", [1, 2, 3, 4, 5, 6], in_place=True)
+    assert "stuff" in tse.col_data.column_names