Use IntegerList and BooleanList for seqinfo's attributes

Author: jkanche

src/genomicranges/SeqInfo.py

@@ -20,7 +20,7 @@ def _validate_seqnames(seqnames):
 
 
 def _validate_seqlengths(seqlengths, num_seqs):
-    if not ut.is_list_of_type(seqlengths, int, ignore_none=True):
+    if not (isinstance(seqlengths, ut.IntegerList) or ut.is_list_of_type(seqlengths, int, ignore_none=True)):
         raise ValueError("'seqlengths' should be a list of integers.")
 
     if num_seqs != len(seqlengths):
@@ -32,7 +32,7 @@ def _validate_seqlengths(seqlengths, num_seqs):
 
 
 def _validate_is_circular(is_circular, num_seqs):
-    if not ut.is_list_of_type(is_circular, bool, ignore_none=True):
+    if not (isinstance(is_circular, ut.BooleanList) or ut.is_list_of_type(is_circular, bool, ignore_none=True)):
         raise ValueError("'is_circular' should be a list of booleans.")
 
     if num_seqs != len(is_circular):
@@ -141,8 +141,8 @@ def __init__(
         """
         self._seqnames = list(seqnames)
         self._reverse_seqnames = None
-        self._seqlengths = self._flatten_incoming(seqlengths, int)
-        self._is_circular = self._flatten_incoming(is_circular, bool)
+        self._seqlengths = ut.IntegerList(self._flatten_incoming(seqlengths, int))
+        self._is_circular = ut.BooleanList(self._flatten_incoming(is_circular, bool))
         self._genome = self._flatten_incoming(genome, str)
 
         if validate:
@@ -359,7 +359,7 @@ def seqnames(self, seqnames: Sequence[str]):
     ######>> seqlengths <<######
     ############################
 
-    def get_seqlengths(self) -> List[int]:
+    def get_seqlengths(self) -> ut.IntegerList:
         """
         Returns:
             A list of integers is returned containing the lengths of all
@@ -399,7 +399,7 @@ def set_seqlengths(
         return output
 
     @property
-    def seqlengths(self) -> List[int]:
+    def seqlengths(self) -> ut.IntegerList:
         warn(
             "'seqlengths' is deprecated, use 'get_seqlengths' instead",
             UserWarning,
@@ -419,7 +419,7 @@ def seqlengths(self, seqlengths: Optional[Union[int, Sequence[int], Dict[str, in
     ######>> is-circular <<######
     #############################
 
-    def get_is_circular(self) -> List[bool]:
+    def get_is_circular(self) -> ut.BooleanList:
         """
         Returns:
             A list of booleans is returned specifying whether each sequence
@@ -459,7 +459,7 @@ def set_is_circular(
         return output
 
     @property
-    def is_circular(self) -> List[bool]:
+    def is_circular(self) -> ut.BooleanList:
         warn(
             "'is_circular' is deprecated, use 'get_is_circular' instead",
             UserWarning,

tests/test_SeqInfo.py

@@ -20,13 +20,13 @@ def test_create_SeqInfo():
     )
 
     assert len(seq) == 3
-    assert seq.get_is_circular() == circ
+    assert list(seq.get_is_circular()) == circ
     assert seq.get_genome() == ["hg19"] * 3
-    assert seq.get_seqlengths() == [100, 101, 102]
+    assert list(seq.get_seqlengths()) == [100, 101, 102]
 
     assert seq.get_seqnames() == ["chr1", "chr2", "chr3"]
     seq2 = seq.set_seqlengths({"chr2": 500, "chr1": 123, "chr3": 99})
-    assert seq2.get_seqlengths() == [123, 500, 99]
+    assert list(seq2.get_seqlengths()) == [123, 500, 99]
 
     seq2 = seq.set_is_circular(False)
     assert seq2.get_is_circular() == [False] * 3
@@ -113,7 +113,7 @@ def test_subset_seqinfo():
 
     assert isinstance(subset, SeqInfo)
     assert subset.seqnames == ["chr2", "chr3"]
-    assert subset.seqlengths == [101, 102]
+    assert list(subset.seqlengths) == [101, 102]
 
 
 def test_merge_SeqInfo():
@@ -139,8 +139,8 @@ def test_merge_SeqInfo():
 
     combined = merge_SeqInfo([seq, seq2])
     assert combined.get_seqnames() == ["chr1", "chr2", "chr3", "chr4", "chr5"]
-    assert combined.get_seqlengths() == [100, 101, None, 101, 102]
-    assert combined.get_is_circular() == [False, True, False, True, False]
+    assert list(combined.get_seqlengths()) == [100, 101, None, 101, 102]
+    assert list(combined.get_is_circular()) == [False, True, False, True, False]
     assert combined.get_genome() == ["hg19", "hg19", None, "hg38", "hg38"]
 
 
@@ -160,3 +160,10 @@ def test_copy_SeqInfo():
 
     deepcopy_seq = deepcopy(seq)
     assert seq.seqnames == deepcopy_seq.seqnames
+
+
+def test_seqlength_types():
+    s = SeqInfo(["A"], seqlengths=[np.uint64(100)])
+
+    assert isinstance(s, SeqInfo)
+    assert list(s._seqlengths) == [100]