v0.7.43

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "toolchest-client"
-version = "0.7.40"
+version = "0.7.43"
 description = "Python client for Toolchest"
 authors = [
     "Bryce Cai <bcai@trytoolchest.com>",

toolchest_client/__init__.py

@@ -21,4 +21,4 @@
 from toolchest_client.api.auth import get_key, set_key
 from toolchest_client.api.exceptions import ToolchestException, DataLimitError, ToolchestJobError
 from toolchest_client.api.query import Query
-from .tools.api import bowtie2, cellranger_mkfastq, kraken2, shi7, shogun_align, shogun_filter, STAR, test, unicycler
+from .tools.api import bowtie2, cellranger_mkfastq, kraken2, megahit, shi7, shogun_align, shogun_filter, STAR, test, unicycler

toolchest_client/tools/__init__.py

@@ -2,6 +2,7 @@
 from .bowtie2 import Bowtie2
 from .cellranger import CellRangerMkfastq
 from .kraken2 import Kraken2
+from .megahit import Megahit
 from .shi7 import Shi7
 from .shogun import ShogunAlign, ShogunFilter
 from .star import STARInstance

toolchest_client/tools/api.py

@@ -5,7 +5,7 @@
 This module contains the API for using Toolchest tools.
 """
 from toolchest_client.files import assert_exists
-from toolchest_client.tools import Kraken2, CellRangerMkfastq, Bowtie2, Shi7, ShogunAlign, ShogunFilter, STARInstance, Test, Unicycler
+from toolchest_client.tools import Kraken2, CellRangerMkfastq, Bowtie2, Megahit, Shi7, ShogunAlign, ShogunFilter, STARInstance, Test, Unicycler
 
 
 def bowtie2(inputs, output_path, database_name="GRCh38_noalt_as", database_version="1", tool_args=""):
@@ -102,7 +102,7 @@ def kraken2(output_path, inputs=[], database_name="standard", database_version="
         >>> toolchest.kraken2(
         ...     tool_args="",
         ...     inputs="./path/to/input.fastq",
-        ...     output_path="./path/to/output.txt",
+        ...     output_path="./path/to/output",
         ... )
 
     """
@@ -135,6 +135,61 @@ def kraken2(output_path, inputs=[], database_name="standard", database_version="
     instance.run()
 
 
+def megahit(output_path, tool_args="", read_one=None, read_two=None, interleaved=None,
+            single_end=None):
+    """Runs Megahit via Toolchest.
+
+    :param output_path: Path (client-side) where the output will be downloaded.
+    :param tool_args: (optional) Additional arguments to be passed to Megahit.
+    :param read_one: (optional) `-1` inputs. Path or list of paths for read 1 of paired-read input files.
+    :param read_two: (optional) `-2` inputs. Path or list of paths for read 2 of paired-read input files.
+    :param interleaved: (optional) `--12` inputs. Path or list of paths for interleaved paired-end files.
+    :param single_end: (optional) `-r` inputs. Path or list of paths for single-end inputs.
+
+    .. note:: Each read in `read_one` should match with a read in `read_two`, and vice
+    versa. In other words, the nth read in `read_one` should be paired with the nth read
+    in `read_two`.
+
+    Usage::
+
+        >>> import toolchest_client as toolchest
+        >>> toolchest.megahit(
+        ...     tool_args="",
+        ...     read_one=["./pair_1/r1.fa", "./pair_2/r1.fa"],
+        ...     read_two=["./pair_1/r2.fa", "./pair_2/r2.fa"],
+        ...     output_path="./path/to/output",
+        ... )
+
+    """
+
+    # If input parameters are lists, parse these for input_prefix_mapping.
+    tag_to_param_map = {
+        "-1": read_one,
+        "-2": read_two,
+        "--12": interleaved,
+        "-r": single_end,
+    }
+    input_list = []  # list of all inputs
+    input_prefix_mapping = {}  # map of each input to its respective tag
+    for tag, param in tag_to_param_map.items():
+        if isinstance(param, list):
+            for input_file in param:
+                input_list.append(input_file)
+                input_prefix_mapping[input_file] = tag
+        elif isinstance(param, str):
+            input_list.append(param)
+            input_prefix_mapping[param] = tag
+
+    instance = Megahit(
+        tool_args=tool_args,
+        output_name='output.tar.gz',
+        input_prefix_mapping=input_prefix_mapping,
+        inputs=input_list,
+        output_path=output_path,
+    )
+    instance.run()
+
+
 def shi7(inputs, output_path, tool_args=""):
     """Runs shi7 via Toolchest.
 
@@ -189,7 +244,7 @@ def shogun_align(inputs, output_path, database_name="shogun_standard", database_
         inputs=inputs,
         output_path=output_path,
         database_name=database_name,
-        database_version=database_version
+        database_version=database_version,
     )
     instance.run()
 
@@ -222,7 +277,7 @@ def shogun_filter(inputs, output_path, database_name="shogun_standard", database
         inputs=inputs,
         output_path=output_path,
         database_name=database_name,
-        database_version=database_version
+        database_version=database_version,
     )
     instance.run()
 

toolchest_client/tools/megahit.py

@@ -0,0 +1,38 @@
+"""
+toolchest_client.tools.megahit
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+This is the megahit implementation of the Tool class.
+"""
+from toolchest_client.files import OutputType
+
+from . import Tool
+
+
+class Megahit(Tool):
+    """
+    The megahit implementation of the Tool class.
+    """
+    def __init__(self, tool_args, output_name, inputs, input_prefix_mapping,
+                 output_path):
+        super().__init__(
+            tool_name="megahit",
+            tool_version="1.2.9",  # todo: allow version to be set by the user
+            tool_args=tool_args,
+            output_name=output_name,
+            output_path=output_path,
+            inputs=inputs,
+            input_prefix_mapping=input_prefix_mapping,
+            min_inputs=1,
+            max_inputs=10,  # todo: make this unlimited?
+            parallel_enabled=False,
+            output_type=OutputType.GZ_TAR,
+            output_is_directory=True,
+            output_names=[
+                "checkpoints.txt",
+                "done",
+                "final.contigs.fa",
+                "log",
+                "options.json",
+            ],
+        )

toolchest_client/tools/tool_args.py

@@ -123,6 +123,32 @@
             "--paired": 0,
         },
     },
+    "megahit": {
+        "blacklist": [
+            "--continue",
+            "-o",
+        ],
+        "whitelist": {
+            "--min-count": 1,
+            "--k-list": 1,
+            "--k-min": 1,
+            "--k-max": 1,
+            "--k-step": 1,
+            "--no-mercy": 0,
+            "--bubble-level": 1,
+            "--merge-level": 1,
+            "--prune-level": 1,
+            "--prune-depth": 1,
+            "--disconnect-ratio": 1,
+            "--low-local-ratio": 1,
+            "--max-tip-len": 1,
+            "--cleaning-rounds": 1,
+            "--no-local": 0,
+            "--kmin-1pass": 1,
+            "--presets": 1,
+            "--min-contig-len": 1,
+        },
+    },
     "shi7": {
         "whitelist": {
             "--debug": 0,