Slice RSE by granges

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## Version 0.7.0
+
+- Extended `__getitem__` subscripting on `RangedSummarizedExperiment` to support direct slicing by `GenomicRanges` or `CompressedGenomicRangesList` objects.
+
 ## Version 0.6.0 - 0.6.5
 
 - Classes now extend `BiocObject` from biocutils, which provides a metadata field.

src/summarizedexperiment/RangedSummarizedExperiment.py

@@ -392,6 +392,17 @@ def seq_info(self) -> SeqInfo:
 
     # rest of them are inherited from BaseSE.
 
+    def _normalize_row_slice(self, rows: Union[str, int, bool, Sequence]):
+
+        if isinstance(rows, (GenomicRanges, CompressedGenomicRangesList)):
+            hits = self.row_ranges.find_overlaps(query=rows)
+            rows = hits.get_column("self_hits")
+        elif hasattr(rows, "find_overlaps"):
+            hits = self.row_ranges.find_overlaps(query=rows)
+            rows = hits.get_column("self_hits")
+
+        return super()._normalize_row_slice(rows)
+
     def get_slice(
         self,
         rows: Optional[Union[str, int, bool, Sequence]],

src/summarizedexperiment/base.py

@@ -33,7 +33,10 @@ def _guess_assay_shape(assays, rows, cols, row_names, col_names) -> tuple:
     _keys = list(assays.keys())
     if len(_keys) > 0:
         _first = _keys[0]
-        return assays[_first].shape
+        mat = assays[_first]
+        if not hasattr(mat, "shape"):
+            raise TypeError(f"Assay: '{_first}' is not a supported matrix representation.")
+        return mat.shape
 
     _r = 0
     if rows is not None:
@@ -201,14 +204,14 @@ def __init__(
         self._cols = _sanitize_frame(column_data, self._shape[1])
 
         if row_names is None:
-            row_names = self._rows.row_names
+            row_names = self._rows.row_names if hasattr(self._rows, "row_names") else None
 
         if row_names is not None and not isinstance(row_names, ut.Names):
             row_names = ut.Names(row_names)
         self._row_names = row_names
 
         if column_names is None:
-            column_names = self._cols.row_names
+            column_names = self._cols.row_names if hasattr(self._cols, "row_names") else None
 
         if column_names is not None and not isinstance(column_names, ut.Names):
             column_names = ut.Names(column_names)
@@ -1125,7 +1128,7 @@ def __getitem__(
         Returns:
             Same type as caller with the sliced rows and columns.
         """
-        if isinstance(args, (str, int)):
+        if not isinstance(args, tuple):
             return self.get_slice(args, slice(None))
 
         if isinstance(args, tuple):
@@ -1139,8 +1142,6 @@ def __getitem__(
             else:
                 raise ValueError(f"`{type(self).__name__}` only supports 2-dimensional slicing.")
 
-        raise TypeError("args must be a sequence or a scalar integer or string or a tuple of atmost 2 values.")
-
     ################################
     ######>> AnnData interop <<#####
     ################################

tests/test_RSE_methods.py

@@ -4,6 +4,9 @@
 import numpy as np
 import pandas as pd
 import pytest
+from genomicranges import GenomicRanges
+from iranges import IRanges
+
 from summarizedexperiment.RangedSummarizedExperiment import RangedSummarizedExperiment
 from summarizedexperiment.SummarizedExperiment import SummarizedExperiment
 
@@ -15,42 +18,36 @@
 nrows = 200
 ncols = 6
 counts = np.random.rand(nrows, ncols)
-df_gr = pd.DataFrame(
-    {
-        "seqnames": [
-            "chr1",
-            "chr2",
-            "chr2",
-            "chr2",
-            "chr1",
-            "chr1",
-            "chr3",
-            "chr3",
-            "chr3",
-            "chr3",
-        ]
-        * 20,
-        "starts": range(100, 300),
-        "ends": range(110, 310),
-        "strand": ["-", "+", "+", "*", "*", "+", "+", "+", "-", "-"] * 20,
-        "score": range(0, 200),
-        "GC": [random() for _ in range(10)] * 20,
-    }
-)
+df_gr = pd.DataFrame({
+    "seqnames": [
+        "chr1",
+        "chr2",
+        "chr2",
+        "chr2",
+        "chr1",
+        "chr1",
+        "chr3",
+        "chr3",
+        "chr3",
+        "chr3",
+    ]
+    * 20,
+    "starts": range(100, 300),
+    "ends": range(110, 310),
+    "strand": ["-", "+", "+", "*", "*", "+", "+", "+", "-", "-"] * 20,
+    "score": range(0, 200),
+    "GC": [random() for _ in range(10)] * 20,
+})
 
 gr = genomicranges.GenomicRanges.from_pandas(df_gr)
 
-col_data = pd.DataFrame(
-    {
-        "treatment": ["ChIP", "Input"] * 3,
-    }
-)
+col_data = pd.DataFrame({
+    "treatment": ["ChIP", "Input"] * 3,
+})
 
 
 def test_RSE_props():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -78,9 +75,7 @@ def test_RSE_props():
 
 
 def test_RSE_subset():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -96,9 +91,7 @@ def test_RSE_subset():
 
 
 def test_RSE_subset_assays():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -112,9 +105,7 @@ def test_RSE_subset_assays():
 
 
 def test_RSE_coverage():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -125,9 +116,7 @@ def test_RSE_coverage():
 
 
 def test_RSE_distance_methods():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -143,9 +132,7 @@ def test_RSE_distance_methods():
 
 
 def test_RSE_range_methods():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -170,9 +157,7 @@ def test_RSE_range_methods():
 
 
 def test_RSE_search():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -185,9 +170,7 @@ def test_RSE_search():
 
 
 def test_RSE_sort_order():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -200,10 +183,9 @@ def test_RSE_sort_order():
     assert res is not None
     assert len(res.row_ranges) == len(tse.row_ranges)
 
+
 def test_RSE_assay_getters_and_setters():
-    tse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     assert tse is not None
     assert isinstance(tse, RangedSummarizedExperiment)
@@ -221,21 +203,19 @@ def test_RSE_assay_getters_and_setters():
     assert tse.get_assay("new_counts") is not None
     assert new_tse.get_assay("new_counts") is not None
 
+
 def test_RSE_to_se():
-    rse = RangedSummarizedExperiment(
-        assays={"counts": counts}, row_ranges=gr, column_data=col_data
-    )
+    rse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     se = rse.to_summarizedexperiment()
 
     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
-    )
+    tse = RangedSummarizedExperiment(assays={"counts": counts}, row_ranges=gr, column_data=col_data)
 
     se = tse.to_summarizedexperiment()
 
@@ -247,3 +227,62 @@ def test_RSE_to_se_with_ranges():
     assert "seqnames" in se.row_data.column_names
     assert "starts" in se.row_data.column_names
     assert "ends" in se.row_data.column_names
+
+
+def test_RSE_combine():
+    gr1 = GenomicRanges(seqnames=["chr1", "chr2"], ranges=IRanges([1, 100], [10, 50]), strand=["+", "-"])
+    rse1 = RangedSummarizedExperiment(
+        assays={"counts": np.array([[1, 2], [3, 4]])}, row_ranges=gr1, column_names=["sample1", "sample2"]
+    )
+
+    gr2 = GenomicRanges(seqnames=["chr3"], ranges=IRanges([200], [20]), strand=["*"])
+    rse2 = RangedSummarizedExperiment(
+        assays={"counts": np.array([[5, 6]])}, row_ranges=gr2, column_names=["sample1", "sample2"]
+    )
+
+    combined_rows = rse1.combine_rows(rse2)
+    assert combined_rows.shape == (3, 2)
+    assert list(combined_rows.row_ranges.get_seqnames()) == ["chr1", "chr2", "chr3"]
+
+    combined_rows_rel = rse1.relaxed_combine_rows(rse2)
+    assert combined_rows_rel.shape == (3, 2)
+
+    rse3 = RangedSummarizedExperiment(assays={"counts": np.array([[7], [8]])}, row_ranges=gr1, column_names=["sample3"])
+    combined_cols = rse1.combine_columns(rse3)
+    assert combined_cols.shape == (2, 3)
+    assert list(combined_cols.column_names) == ["sample1", "sample2", "sample3"]
+
+    combined_cols_rel = rse1.relaxed_combine_columns(rse3)
+    assert combined_cols_rel.shape == (2, 3)
+
+
+def test_RSE_genomic_slice():
+    df_gr = pd.DataFrame({
+        "seqnames": ["chr1", "chr2", "chr3"],
+        "starts": [100, 200, 300],
+        "ends": [150, 250, 350],
+        "strand": ["+", "-", "*"],
+    })
+    gr = genomicranges.GenomicRanges.from_pandas(df_gr)
+
+    rse = RangedSummarizedExperiment(
+        assays={"counts": np.random.rand(3, 2)},
+        row_ranges=gr,
+    )
+
+    query_df = pd.DataFrame({
+        "seqnames": ["chr2"],
+        "starts": [220],
+        "ends": [230],
+        "strand": ["-"],
+    })
+    query_gr = genomicranges.GenomicRanges.from_pandas(query_df)
+
+    sliced = rse[query_gr, :]
+    assert sliced is not None
+    assert sliced.shape == (1, 2)
+    assert sliced.row_ranges.get_seqnames() == ["chr2"]
+
+    sliced_1d = rse[query_gr]
+    assert sliced_1d.shape == (1, 2)
+    assert sliced_1d.row_ranges.get_seqnames() == ["chr2"]