From da6ccad6eefb864fd48d8376069a2adac5ba8993 Mon Sep 17 00:00:00 2001
From: ChocoParrot <lachocoparrot@gmail.com>
Date: Mon, 13 Sep 2021 13:25:22 +0800
Subject: [PATCH] [factor] GTF command line tool

---
 setup.py                            |  3 ++
 src/cline_tools/__init__.py         |  1 +
 src/cline_tools/orffinder-to-gtf.py | 55 +++++++++++++++++++++++++++++
 3 files changed, 59 insertions(+)
 create mode 100644 src/cline_tools/__init__.py
 create mode 100644 src/cline_tools/orffinder-to-gtf.py

diff --git a/setup.py b/setup.py
index 2226273..7212f43 100644
--- a/setup.py
+++ b/setup.py
@@ -22,6 +22,9 @@
         "License :: OSI Approved :: MIT License",
         "Operating System :: OS Independent",
     ),
+    install_requires=[
+        "biopython=1.79"
+    ]
     package_dir={"": "src"},
     packages=setuptools.find_packages(where="src"),
     python_requires=">=3.6"
diff --git a/src/cline_tools/__init__.py b/src/cline_tools/__init__.py
new file mode 100644
index 0000000..291e01a
--- /dev/null
+++ b/src/cline_tools/__init__.py
@@ -0,0 +1 @@
+__name__ = "orffinder"
diff --git a/src/cline_tools/orffinder-to-gtf.py b/src/cline_tools/orffinder-to-gtf.py
new file mode 100644
index 0000000..bb8cd80
--- /dev/null
+++ b/src/cline_tools/orffinder-to-gtf.py
@@ -0,0 +1,55 @@
+import sys
+from Bio import SeqIO
+from orffinder import orffinder
+
+arguments = sys.argv
+classed_arguments = {"orf_size": "75", "max_orfs_per_sequence": "-1", "remove_nested": "False", "trim_trailing": "False", "infmt": "fasta", "attr_name": "ORF_"}
+
+try:
+    for i in range(len(arguments)):
+
+        argument = arguments[i]
+
+        if argument.startswith("-"):
+
+            classed_arguments[argument[1:]] = arguments[i + 1]
+
+    sequences = SeqIO.parse(classed_arguments["in"], classed_arguments["infmt"])
+
+    orf_size = int(classed_arguments["orf_size"])
+    remove_nested = classed_arguments["remove_nested"] == "True"
+    trim_trailing = classed_arguments["trim_trailing"] == "True"
+    attr_name = classed_arguments["attr_name"]
+    max_orfs_per_sequence = int(classed_arguments["max_orfs_per_sequence"])
+
+    output = list()
+    index = int()
+
+    for sequence in sequences:
+
+        seqname = sequence.description
+        orfs = orffinder.getORFs(sequence, minimum_length=orf_size, trim_trailing=trim_trailing, remove_nested=remove_nested)
+
+        local_index = int()
+        for orf in orfs:
+
+            index += 1
+            local_index += 1
+            output.append([seqname, "ORFFinder Python", "ORF", str(orf["start"]), str(orf["end"]), ".", orf["sense"], str(orf["frame"] - 1), "orf_id \"" + attr_name + str(index) + "\""])
+
+            if local_index >= max_orfs_per_sequence and max_orfs_per_sequence != -1:
+                break
+
+
+    full_output = "\n".join(["\t".join(x) for x in output])
+
+    if "out" not in classed_arguments.keys():
+
+        print(full_output)
+
+    else:
+
+        open(classed_arguments["out"], "w+").write(full_output)
+
+except:
+    print("USAGE\n  orffinder-to-gtf [-in input] [-infmt format] [-out output] [-orf_size int]\n    [-remove_nested boolean] [-trim_trailing boolean] [-max_orfs_per_sequence int]\n    [-attr_name string]\n\nDESCRIPTION\n  ORFFinder Python v1.5\n\nUse '-help' to print detailed descriptions of command line arguments\n========================================================================")