biopipen 0.31.6__tar.gz → 0.31.7__tar.gz

{biopipen-0.31.6 → biopipen-0.31.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: biopipen
-Version: 0.31.6
+Version: 0.31.7
 Summary: Bioinformatics processes/pipelines that can be run from `pipen run`
 License: MIT
 Author: pwwang

biopipen-0.31.7/biopipen/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.31.7"

{biopipen-0.31.6 → biopipen-0.31.7}/biopipen/ns/bam.py

@@ -301,3 +301,31 @@ class BamSampling(Proc):
         "sort_args": [],
     }
     script = "file://../scripts/bam/BamSampling.py"
+
+
+class BamSubsetByBed(Proc):
+    """Subset bam file by the regions in a bed file
+
+    Input:
+        bamfile: The bam file
+        bedfile: The bed file
+
+    Output:
+        outfile: The output bam file
+
+    Envs:
+        ncores: Number of cores to use
+        samtools: Path to samtools executable
+        tool: The tool to use, currently only "samtools" is supported
+        index: Whether to index the output bam file
+    """
+    input = "bamfile:file, bedfile:file"
+    output = "outfile:file:{{in.bamfile | stem}}-subset.bam"
+    lang = config.lang.python
+    envs = {
+        "ncores": config.misc.ncores,
+        "samtools": config.exe.samtools,
+        "tool": "samtools",
+        "index": True,
+    }
+    script = "file://../scripts/bam/BamSubsetByBed.py"

{biopipen-0.31.6 → biopipen-0.31.7}/biopipen/ns/bed.py

@@ -198,3 +198,43 @@ class BedtoolsIntersect(Proc):
         "postcmd": None,
     }
     script = "file://../scripts/bed/BedtoolsIntersect.py"
+
+
+class BedtoolsMakeWindows(Proc):
+    """Make windows from a BED file or genome size file, using `bedtools makewindows`.
+
+    Input:
+        infile: The input BED file or a genome size file
+            Type will be detected by the number of columns in the file.
+            If it has 3+ columns, it is treated as a BED file, otherwise
+            a genome size file.
+
+    Output:
+        outfile: The output BED file
+
+    Envs:
+        bedtools: The path to bedtools
+        window (type=int): The size of the windows
+        step (type=int): The step size of the windows
+        nwin (type=int): The number of windows to be generated
+            Exclusive with `window` and `step`.
+            Either `nwin` or `window` and `step` should be provided.
+        reverse (flag): Reverse numbering of windows in the output
+        name (choice): How to name the generated windows/regions
+            - none: Do not add any name
+            - src: Use the source interval's name
+            - winnum: Use the window number
+            - srcwinnum: Use the source interval's name and window number
+    """  # noqa: E501
+    input = "infile:file"
+    output = "outfile:file:{{in.infile | stem}}_windows.bed"
+    lang = config.lang.python
+    envs = {
+        "bedtools": config.exe.bedtools,
+        "window": None,
+        "step": None,
+        "nwin": None,
+        "reverse": False,
+        "name": "none",
+    }
+    script = "file://../scripts/bed/BedtoolsMakeWindows.py"

biopipen-0.31.7/biopipen/scripts/bam/BamSubsetByBed.py

@@ -0,0 +1,38 @@
+from pathlib import Path
+from biopipen.utils.misc import run_command, logger
+
+# using:
+# samtools view --subsample 0.1 --subsample-seed 1234 --threads 4 -b -o out.bam in.bam
+
+bamfile = {{ in.bamfile | repr }} # pyright: ignore # noqa
+bedfile = {{ in.bedfile | repr }} # pyright: ignore # noqa
+outfile = Path({{ out.outfile | repr }}) # pyright: ignore
+ncores = {{ envs.ncores | int }} # pyright: ignore
+samtools = {{ envs.samtools | repr }} # pyright: ignore
+tool = {{ envs.tool | repr }} # pyright: ignore
+should_index = {{ envs.index | repr }} # pyright: ignore
+
+if tool != "samtools":
+    raise ValueError(
+        f"Tool {tool} is not supported. "
+        "Currently only samtools is supported."
+    )
+
+cmd = [
+    samtools,
+    "view",
+    "--target-file",
+    bedfile,
+    "-b",
+    "--threads",
+    ncores,
+    "-o",
+    outfile,
+    bamfile
+]
+run_command(cmd, fg=True)
+
+if should_index:
+    logger.info("Indexing the output bam file.")
+    cmd = [samtools, "index", "-@", ncores, outfile]
+    run_command(cmd, fg=True)

biopipen-0.31.7/biopipen/scripts/bed/BedtoolsMakeWindows.py

@@ -0,0 +1,47 @@
+from pathlib import Path
+from biopipen.utils.misc import run_command, logger
+
+infile = Path({{in.afile | repr}})  # pyright: ignore # noqa: #999
+outfile = Path({{in.bfile | repr}})  # pyright: ignore
+bedtools = {{envs.bedtools | repr}}  # pyright: ignore
+window = {{envs.window | repr}}  # pyright: ignore
+step = {{envs.step | repr}}  # pyright: ignore
+nwin = {{envs.nwin | repr}}  # pyright: ignore
+reverse = {{envs.reverse | repr}}  # pyright: ignore
+name = {{envs.name | repr}}  # pyright: ignore
+
+if nwin is None and window is None:
+    raise ValueError("Either `nwin` or `window` should be provided.")
+
+if nwin is not None and window is not None:
+    raise ValueError("Either `nwin` or `window` should be provided, not both.")
+
+# detect if infile is a genome size file or a bed file
+with infile.open() as f:
+    line = f.readline().strip()
+    if len(line.split("\t")) > 2:
+        is_bed = True
+    else:
+        is_bed = False
+
+if is_bed:
+    logger.info("BED file is detected as input.")
+    cmd = [bedtools, "makewindows", "-b", infile]
+else:
+    logger.info("Genome size file is detected as input.")
+    cmd = [bedtools, "makewindows", "-g", infile]
+
+if nwin:
+    cmd.extend(["-n", nwin])
+elif step is not None:
+    cmd.extend(["-w", window, "-s", step])
+else:
+    cmd.extend(["-w", window])
+
+if reverse:
+    cmd.append("-reverse")
+
+if name != "none":
+    cmd.extend(["-name", name])
+
+run_command(cmd, stdout=outfile)

{biopipen-0.31.6 → biopipen-0.31.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "biopipen"
-version = "0.31.6"
+version = "0.31.7"
 description = "Bioinformatics processes/pipelines that can be run from `pipen run`"
 authors = ["pwwang <pwwang@pwwang.com>"]
 license = "MIT"
@@ -58,3 +58,6 @@ build-backend = "poetry.core.masonry.api"
 line-length = 88
 target-version = ['py39', 'py310', 'py311', 'py312']
 include = '\.pyi?$'
+
+[tool.pyright]
+include = [ "biopipen/**/*.py" ]

{biopipen-0.31.6 → biopipen-0.31.7}/setup.py

@@ -85,7 +85,7 @@ entry_points = \
 
 setup_kwargs = {
     'name': 'biopipen',
-    'version': '0.31.6',
+    'version': '0.31.7',
     'description': 'Bioinformatics processes/pipelines that can be run from `pipen run`',
     'long_description': 'None',
     'author': 'pwwang',

biopipen-0.31.6/biopipen/__init__.py

@@ -1 +0,0 @@
-__version__ = "0.31.6"