Merge pull request #6 from alienzj/modular

Modular

MANIFEST.in

@@ -0,0 +1,15 @@
+include *.md
+include LICENSE
+
+include bin/graphbin
+
+recursive-include graphbin/ *
+recursive-include graphbin/bidirectionalmap *
+recursive-include graphbin/labelpropagation *
+
+global-exclude graphbin/ *.pyc
+global-exclude graphbin/__pycache__ *.pyc
+global-exclude graphbin/bidirectionalmap *.pyc
+global-exclude graphbin/bidirectionalmap/__pycache__ *.pyc
+global-exclude graphbin/labelpropagation *.pyc
+global-exclude graphbin/labelpropagation/__pycache__ *.pyc

bin/graphbin

@@ -0,0 +1,132 @@
+#!/usr/bin/env python3
+
+"""graphbin: Refined binning of metagenomic contigs using assembly graphs."""
+
+import argparse
+import os
+import sys
+
+from graphbin import (
+    graphbin_Options,
+    graphbin_Canu,
+    graphbin_Flye,
+    graphbin_MEGAHIT,
+    graphbin_Miniasm,
+    graphbin_SGA,
+    graphbin_SPAdes,
+)
+
+__author__ = "Vijini Mallawaarachchi, Anuradha Wickramarachchi, and Yu Lin"
+__copyright__ = "Copyright 2020, GraphBin Project"
+__credits__ = "Benjamin Kaehler and Gavin Huttley"
+__license__ = "GPL"
+__version__ = "1.1"
+__maintainer__ = "Vijini Mallawaarachchi"
+__email__ = "[email protected]"
+__status__ = "Stable Release"
+
+
+def run(args):
+    RUNNER = {
+        "canu": graphbin_Canu.run,
+        "flye": graphbin_Flye.run,
+        "megahit": graphbin_MEGAHIT.run,
+        "miniasm": graphbin_Miniasm.run,
+        "sga": graphbin_SGA.run,
+        "spades": graphbin_SPAdes.run,
+    }
+    RUNNER[args.assembler](args)
+
+
+def main():
+    parser = graphbin_Options.PARSER
+    parser.add_argument(
+        "--assembler",
+        type=str,
+        help="name of the assembler used (SPAdes, SGA or MEGAHIT). GraphBin supports Flye, Canu and Miniasm long-read assemblies as well.",
+    )
+    parser.add_argument(
+        "--paths",
+        default=None,
+        required=False,
+        help="path to the contigs.paths file, only SPAdes need",
+    )
+
+    args = parser.parse_args()
+
+    if args.version:
+        print("GraphBin version %s" % __version__)
+        sys.exit(0)
+
+    # Validation of inputs
+    # ---------------------------------------------------
+    # Check assembler type
+    if not (
+        args.assembler.lower() == "spades"
+        or args.assembler.lower() == "sga"
+        or args.assembler.lower() == "megahit"
+        or args.assembler.lower() == "flye"
+        or args.assembler.lower() == "canu"
+        or args.assembler.lower() == "miniasm"
+    ):
+        print(
+            "\nPlease make sure to provide the correct assembler type (SPAdes, SGA or MEGAHIT). GraphBin supports Flye, Canu and Miniasm long-read assemblies as well."
+        )
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Check assembly graph file
+    if not os.path.isfile(args.graph):
+        print("\nFailed to open the assembly graph file.")
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Check if paths files is provided when the assembler type is SPAdes
+    if args.assembler.lower() == "spades" and args.paths is None:
+        print("\nPlease make sure to provide the path to the contigs.paths file.")
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Check contigs.paths file for SPAdes
+    if args.assembler.lower() == "spades" and not os.path.isfile(args.paths):
+        print("\nFailed to open the contigs.paths file.")
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Check the file with the initial binning output
+    if not os.path.isfile(args.binned):
+        print("\nFailed to open the file with the initial binning output.")
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Handle for missing trailing forwardslash in output folder path
+    if args.output[-1:] != "/":
+        args.output = args.output + "/"
+
+    # Create output folder if it does not exist
+    os.makedirs(args.output, exist_ok=True)
+
+    # Validate prefix
+    if args.prefix != "":
+        if not args.prefix.endswith("_"):
+            args.prefix = args.prefix + "_"
+
+    # Validate max_iteration
+    if args.max_iteration <= 0:
+        print("\nPlease enter a valid number for max_iteration")
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Validate diff_threshold
+    if args.diff_threshold < 0:
+        print("\nPlease enter a valid number for diff_threshold")
+        print("Exiting GraphBin...\nBye...!\n")
+        sys.exit(1)
+
+    # Run GraphBin
+    # ---------------------------------------------------
+    run(args)
+
+
+if __name__ == "__main__":
+    main()