cutseq 0.0.6__tar.gz → 0.0.8__tar.gz

{cutseq-0.0.6 → cutseq-0.0.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cutseq
-Version: 0.0.6
+Version: 0.0.8
 Summary: Automatically cut adapter / barcode / UMI from NGS data
 Home-page: https://github.com/y9c/cutseq
 License: MIT

{cutseq-0.0.6 → cutseq-0.0.8}/cutseq/run.py

@@ -27,6 +27,7 @@ from cutadapt.modifiers import (
     PairedEndRenamer,
     QualityTrimmer,
     Renamer,
+    SingleEndModifier,
     SuffixRemover,
     UnconditionalCutter,
 )
@@ -55,6 +56,38 @@ def patched_problematic_method(self, *args, **kwargs):
 
 Statistics._collect_modifier = patched_problematic_method
 
+## Enhnace function for cutadapt
+
+
+class IsUntrimmedAny(Predicate):
+    """
+    Select reads for which no adapter match was found
+    """
+
+    def __init__(self, ref_adapters):
+        self.ref_adapters = ref_adapters
+
+    def __repr__(self):
+        return "IsUntrimmedAny()"
+
+    def test(self, read, info):
+        # check not all adapters with ref_adapters are exist in the matches
+        match_adapters = [match.adapter for match in info.matches]
+        if any([adapter not in match_adapters for adapter in self.ref_adapters]):
+            return True
+        return False
+
+
+class ReverseComplementConverter(SingleEndModifier):
+    def __init__(self):
+        self.rcd = False
+
+    def __repr__(self):
+        return "ReverseComplementConverter()"
+
+    def __call__(self, read, info):
+        return read.reverse_complement()
+
 
 __version__ = importlib.metadata.version(__package__ or __name__)
 
@@ -143,25 +176,6 @@ class CutadaptConfig:
         self.threads = 1
 
 
-class IsUntrimmedAny(Predicate):
-    """
-    Select reads for which no adapter match was found
-    """
-
-    def __init__(self, ref_adapters):
-        self.ref_adapters = ref_adapters
-
-    def __repr__(self):
-        return "IsUntrimmedAny()"
-
-    def test(self, read, info):
-        # check not all adapters with ref_adapters are exist in the matches
-        match_adapters = [match.adapter for match in info.matches]
-        if any([adapter not in match_adapters for adapter in self.ref_adapters]):
-            return True
-        return False
-
-
 def run_steps(steps, dry_run=False):
     if dry_run:
         print(
@@ -240,6 +254,12 @@ def pipeline_single(input1, output1, short1, untrimmed1, barcode, settings):
         QualityTrimmer(cutoff_front=0, cutoff_back=settings.min_quality),
     )
 
+    # step 9: reverse complement the read
+    if settings.reverse_complement:
+        if barcode.strand == "+":
+            logging.warn("Library is + strand, but reverse complement is enabled.")
+        modifiers.append(ReverseComplementConverter())
+
     inpaths = InputPaths(input1)
 
     with make_runner(inpaths, cores=settings.threads) as runner:
@@ -249,14 +269,16 @@ def pipeline_single(input1, output1, short1, untrimmed1, barcode, settings):
             interleaved=False,
         )
         steps = []
+        # TODO: report info
+        # --info-file=info.txt
+        # PairedSingleEndStep(InfoFileWriter(outfiles.open_text("info.txt"))),
         steps.append(
-            # --info-file=info.txt
-            # PairedSingleEndStep(InfoFileWriter(outfiles.open_text("info.txt"))),
             # -m 10
             SingleEndFilter(
                 TooShort(settings.min_length), outfiles.open_record_writer(short1)
             ),
         )
+        # TODO: --max-n=0 support
         if (
             settings.ensure_inline_barcode
             and barcode.inline5.len + barcode.inline3.len > 0
@@ -267,7 +289,6 @@ def pipeline_single(input1, output1, short1, untrimmed1, barcode, settings):
             if barcode.inline3.len > 0:
                 ref_adapters.append(adapter_inline3)
             steps.append(
-                # TODO: --max-n=0 support
                 SingleEndFilter(
                     IsUntrimmedAny(ref_adapters),
                     outfiles.open_record_writer(untrimmed1, interleaved=False),
@@ -424,6 +445,12 @@ def pipeline_paired(
         )
     )
 
+    # step 9: reverse complement the read
+    if settings.reverse_complement:
+        if barcode.strand == "+":
+            logging.warn("Library is + strand, but reverse complement is enabled.")
+        modifiers.append((ReverseComplementConverter(), ReverseComplementConverter()))
+
     inpaths = InputPaths(input1, input2)
 
     with make_runner(inpaths, cores=settings.threads) as runner:
@@ -443,12 +470,12 @@ def pipeline_paired(
                 outfiles.open_record_writer(short1, short2, interleaved=False),
             )
         )
+        # TODO: --max-n=0 support
         if (
             settings.ensure_inline_barcode
             and barcode.inline5.len + barcode.inline3.len > 0
         ):
             steps.append(
-                # TODO: --max-n=0 support
                 PairedEndFilter(
                     IsUntrimmedAny([adapter_inline5])
                     if barcode.inline5.len > 0
@@ -473,7 +500,7 @@ def pipeline_paired(
 
 
 def run_cutseq(args):
-    barcode_config = BarcodeConfig(args.adapter_scheme.upper())
+    barcode_config = BarcodeConfig(args.replace(" ", "").adapter_scheme.upper())
     settings = CutadaptConfig()
     settings.rname_suffix = args.with_rname_suffix
     settings.ensure_inline_barcode = args.ensure_inline_barcode
@@ -481,6 +508,7 @@ def run_cutseq(args):
     settings.threads = args.threads
     settings.min_length = args.min_length
     settings.dry_run = args.dry_run
+    settings.reverse_complement = args.reverse_complement
     if len(args.input_file) == 1:
         pipeline_single(
             args.input_file[0],
@@ -583,6 +611,12 @@ def main():
 
     parser.add_argument("--trim-polyA", action="store_true", help="Trim polyA tail.")
 
+    parser.add_argument(
+        "--reverse-complement",
+        action="store_true",
+        help="Reverse complement the reads.",
+    )
+
     parser.add_argument(
         "-t",
         "--threads",
@@ -601,6 +635,11 @@ def main():
         "-V", "--version", action="version", version=f"%(prog)s {__version__}"
     )
 
+    # Check if no arguments were provided
+    if len(sys.argv) == 1:
+        parser.print_help(sys.stdout)
+        sys.exit(0)
+
     args = parser.parse_args()
 
     if args.adapter_name is not None:

{cutseq-0.0.6 → cutseq-0.0.8}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "cutseq"
-version = "0.0.6"
+version = "0.0.8"
 description = "Automatically cut adapter / barcode / UMI from NGS data"
 authors = ["Ye Chang <yech1990@gmail.com>"]
 license = "MIT"