Feature/ samtools typehinted wrappers

ci/check.sh

@@ -44,6 +44,7 @@ if [[ "$1" == "--check" ]]; then
 fi
 
 banner "Executing in conda environment ${CONDA_DEFAULT_ENV} in directory ${root}"
+run "Checking poetry" "poetry check"
 run "Unit Tests"     "python -m pytest -vv -r sx fgpyo"
 run "Import Sorting" "isort --force-single-line-imports --profile black fgpyo"
 run "Style Checking" "black --line-length 99 $black_extra_args fgpyo"

fgpyo/io/tests/test_io.py

@@ -86,8 +86,8 @@ def test_assert_path_is_writeable_pass() -> None:
 @pytest.mark.parametrize(
     "suffix, expected",
     [
-        (".gz", io._io.TextIOWrapper),
-        (".fa", io._io.TextIOWrapper),
+        (".gz", io.TextIOWrapper),
+        (".fa", io.TextIOWrapper),
     ],
 )
 def test_reader(
@@ -103,8 +103,8 @@ def test_reader(
 @pytest.mark.parametrize(
     "suffix, expected",
     [
-        (".gz", io._io.TextIOWrapper),
-        (".fa", io._io.TextIOWrapper),
+        (".gz", io.TextIOWrapper),
+        (".fa", io.TextIOWrapper),
     ],
 )
 def test_writer(

fgpyo/sam/__init__.py

@@ -198,13 +198,14 @@ class SamFileType(enum.Enum):
         ext (str): The standard file extension for this file type.
     """
 
-    def __init__(self, mode: str, ext: str) -> None:
+    def __init__(self, mode: str, ext: str, index_ext: Optional[str]) -> None:
         self.mode = mode
         self.extension = ext
+        self.index_extension = index_ext
 
-    SAM = ("", ".sam")
-    BAM = ("b", ".bam")
-    CRAM = ("c", ".cram")
+    SAM = ("", ".sam", None)
+    BAM = ("b", ".bam", ".bai")
+    CRAM = ("c", ".cram", ".crai")
 
     @classmethod
     def from_path(cls, path: Union[Path, str]) -> "SamFileType":
@@ -911,3 +912,5 @@ class SamOrder(enum.Enum):
     Coordinate = "coordinate"  #: coordinate sorted
     QueryName = "queryname"  #: queryname sorted
     Unknown = "unknown"  # Unknown SAM / BAM / CRAM sort order
+    # Sort by template-coordinate, SO is unsorted, GO is query, SS is template-coordinate:
+    TemplateCoordinate = "unsorted"

fgpyo/sam/builder.py

@@ -27,6 +27,7 @@
 from pysam import AlignmentHeader
 
 from fgpyo import sam
+from fgpyo import samtools
 from fgpyo.sam import SamOrder
 
 
@@ -136,6 +137,11 @@ def __init__(
             "SQ": (sd if sd is not None else SamBuilder.default_sd()),
             "RG": [(rg if rg is not None else SamBuilder.default_rg())],
         }
+        if sort_order == SamOrder.TemplateCoordinate:
+            self._header["HD"] = {
+                **self._header["HD"],
+                **{"GO": "query", "SS": "template-coordinate"},
+            }
         if extra_header is not None:
             self._header = {**self._header, **extra_header}
         self._samheader = AlignmentHeader.from_dict(self._header)
@@ -588,18 +594,14 @@ def to_path(
                     if pred(rec):
                         writer.write(rec)
 
-            samtools_sort_args = ["-o", str(path), fp.name]
-
             file_handle.close()
-            if self._sort_order == SamOrder.QueryName:
-                # Ignore type hints for now until we have wrappers to use here.
-                pysam.sort("-n", *samtools_sort_args)  # type: ignore
-            elif self._sort_order == SamOrder.Coordinate:
-                # Ignore type hints for now until we have wrappers to use here.
-                if index:
-                    samtools_sort_args.insert(0, "--write-index")
-                pysam.sort(*samtools_sort_args)  # type: ignore
-
+            if self._sort_order not in {SamOrder.Unsorted, SamOrder.Unknown}:
+                samtools.sort(
+                    alignment_file=Path(fp.name),
+                    output=path,
+                    index_output=index,
+                    sort_order=self._sort_order,
+                )
         return path
 
     def __len__(self) -> int: