diff options
| -rw-r--r-- | gn3/api/rqtl.py | 44 | ||||
| -rw-r--r-- | gn3/app.py | 2 | ||||
| -rw-r--r-- | gn3/commands.py | 14 | ||||
| -rw-r--r-- | gn3/computations/rqtl.py | 40 | ||||
| -rw-r--r-- | gn3/settings.py | 1 | ||||
| -rw-r--r-- | tests/unit/computations/test_rqtl.py | 41 | ||||
| -rw-r--r-- | tests/unit/test_commands.py | 26 |
7 files changed, 168 insertions, 0 deletions
diff --git a/gn3/api/rqtl.py b/gn3/api/rqtl.py new file mode 100644 index 0000000..de620f7 --- /dev/null +++ b/gn3/api/rqtl.py @@ -0,0 +1,44 @@ +"""Endpoints for running the rqtl cmd""" +from flask import Blueprint +from flask import current_app +from flask import jsonify +from flask import request + +from gn3.computations.rqtl import generate_rqtl_cmd +from gn3.computations.gemma import do_paths_exist + +rqtl = Blueprint("rqtl", __name__) + +@rqtl.route("/compute", methods=["POST"]) +def compute(): + """Given at least a geno_file and pheno_file, generate and +run the rqtl_wrapper script and return the results as JSON + + """ + genofile = request.form['geno_file'] + phenofile = request.form['pheno_file'] + + if not do_paths_exist([genofile, phenofile]): + raise FileNotFoundError + + # Split kwargs by those with values and boolean ones that just convert to True/False + kwargs = ["model", "method", "nperm", "scale", "control_marker"] + boolean_kwargs = ["addcovar", "interval"] + all_kwargs = kwargs + boolean_kwargs + + rqtl_kwargs = {"geno": genofile, "pheno": phenofile} + rqtl_bool_kwargs = [] + for kwarg in all_kwargs: + if kwarg in request.form: + if kwarg in kwargs: + rqtl_kwargs[kwarg] = request.form[kwarg] + if kwarg in boolean_kwargs: + rqtl_bool_kwargs.append(kwarg) + + results = generate_rqtl_cmd( + rqtl_wrapper_cmd=current_app.config.get("RQTL_WRAPPER_CMD"), + rqtl_wrapper_kwargs=rqtl_kwargs, + rqtl_wrapper_bool_kwargs=boolean_kwargs + ) + + return jsonify(results) @@ -5,6 +5,7 @@ from typing import Dict from typing import Union from flask import Flask from gn3.api.gemma import gemma +from gn3.api.rqtl import rqtl from gn3.api.general import general from gn3.api.correlation import correlation from gn3.api.data_entry import data_entry @@ -28,6 +29,7 @@ def create_app(config: Union[Dict, str, None] = None) -> Flask: app.config.from_pyfile(config) app.register_blueprint(general, url_prefix="/api/") app.register_blueprint(gemma, url_prefix="/api/gemma") + app.register_blueprint(rqtl, url_prefix="/api/rqtl") app.register_blueprint(correlation, url_prefix="/api/correlation") app.register_blueprint(data_entry, url_prefix="/api/dataentry") return app diff --git a/gn3/commands.py b/gn3/commands.py index 4b0d62d..255ea1d 100644 --- a/gn3/commands.py +++ b/gn3/commands.py @@ -30,6 +30,20 @@ def compose_gemma_cmd(gemma_wrapper_cmd: str = "gemma-wrapper", cmd += " ".join([f"{arg}" for arg in gemma_args]) return cmd +def compose_rqtl_cmd(rqtl_wrapper_cmd: str, + rqtl_wrapper_kwargs: Dict, + rqtl_wrapper_bool_kwargs: list) -> str: + """Compose a valid R/qtl command given the correct input""" + # Add kwargs with values + cmd = rqtl_wrapper_cmd + " " + " ".join( + [f"--{key} {val}" for key, val in rqtl_wrapper_kwargs.items()]) + + # Add boolean kwargs (kwargs without values) + if rqtl_wrapper_bool_kwargs: + cmd += " " + cmd += " ".join([f"--{val}" for val in rqtl_wrapper_bool_kwargs]) + + return cmd def queue_cmd(conn: Redis, job_queue: str, diff --git a/gn3/computations/rqtl.py b/gn3/computations/rqtl.py new file mode 100644 index 0000000..605e0e1 --- /dev/null +++ b/gn3/computations/rqtl.py @@ -0,0 +1,40 @@ +"""Procedures related rqtl computations""" + +from typing import Dict +from gn3.commands import compose_rqtl_cmd +from gn3.computations.gemma import generate_hash_of_string +from gn3.fs_helpers import get_hash_of_files + +def generate_rqtl_cmd(rqtl_wrapper_cmd: str, + rqtl_wrapper_kwargs: Dict, + rqtl_wrapper_bool_kwargs: list) -> Dict: + """Given the base rqtl_wrapper command and +dict of keyword arguments, return the full rqtl_wrapper command and an +output filename generated from a hash of the genotype and phenotype files + + """ + + # Generate a hash from contents of the genotype and phenotype files + _hash = get_hash_of_files( + [v for k, v in rqtl_wrapper_kwargs.items() if k in ["g", "p"]]) + + # Append to hash a hash of keyword arguments + _hash += generate_hash_of_string( + ",".join([f"{k}:{v}" for k, v in rqtl_wrapper_kwargs.items() if k not in ["g", "p"]])) + + # Append to hash a hash of boolean keyword arguments + _hash += generate_hash_of_string( + ",".join(rqtl_wrapper_bool_kwargs)) + + # Temporarily substitute forward-slashes in hash with underscores + _hash = _hash.replace("/", "_") + + _output_filename = f"{_hash}-output.json" + return { + "output_file": + _output_filename, + "rqtl_cmd": + compose_rqtl_cmd(rqtl_wrapper_cmd=rqtl_wrapper_cmd, + rqtl_wrapper_kwargs=rqtl_wrapper_kwargs, + rqtl_wrapper_bool_kwargs=rqtl_wrapper_bool_kwargs) + } diff --git a/gn3/settings.py b/gn3/settings.py index 2057ce1..ecfd502 100644 --- a/gn3/settings.py +++ b/gn3/settings.py @@ -6,6 +6,7 @@ import os BCRYPT_SALT = "$2b$12$mxLvu9XRLlIaaSeDxt8Sle" # Change this! DATA_DIR = "" GEMMA_WRAPPER_CMD = os.environ.get("GEMMA_WRAPPER", "gemma-wrapper") +RQTL_WRAPPER_CMD = os.environ.get("RQTL_WRAPPER") CACHEDIR = "" REDIS_URI = "redis://localhost:6379/0" REDIS_JOB_QUEUE = "GN3::job-queue" diff --git a/tests/unit/computations/test_rqtl.py b/tests/unit/computations/test_rqtl.py new file mode 100644 index 0000000..b16f136 --- /dev/null +++ b/tests/unit/computations/test_rqtl.py @@ -0,0 +1,41 @@ +"""Test cases for procedures defined in computations.rqtl"""
+import unittest
+
+from unittest import mock
+from gn3.computations.rqtl import generate_rqtl_cmd
+
+class TestRqtl(unittest.TestCase):
+ """Test cases for computations.rqtl module"""
+ @mock.patch("gn3.computations.rqtl.generate_hash_of_string")
+ @mock.patch("gn3.computations.rqtl.get_hash_of_files")
+ def test_generate_rqtl_command(self, mock_get_hash_files, mock_generate_hash_string):
+ """Test computing mapping results with R/qtl"""
+ mock_get_hash_files.return_value = "my-hash1"
+ mock_generate_hash_string.return_value = "my-hash2"
+
+ self.assertEqual(
+ generate_rqtl_cmd(rqtl_wrapper_cmd="rqtl-wrapper",
+ rqtl_wrapper_kwargs={
+ "g": "genofile",
+ "p": "phenofile",
+ "model": "normal",
+ "method": "hk",
+ "nperm": 1000,
+ "scale": "Mb",
+ "control": "rs123456"
+ },
+ rqtl_wrapper_bool_kwargs=[
+ "addcovar",
+ "interval"
+ ]), {
+ "output_file":
+ "my-hash1my-hash2my-hash2-output.json",
+ "rqtl_cmd": (
+ "rqtl-wrapper "
+ "--g genofile --p phenofile "
+ "--model normal --method hk "
+ "--nperm 1000 --scale Mb "
+ "--control rs123456 "
+ "--addcovar --interval"
+ )
+ })
diff --git a/tests/unit/test_commands.py b/tests/unit/test_commands.py index aafb3a2..a3d0273 100644 --- a/tests/unit/test_commands.py +++ b/tests/unit/test_commands.py @@ -6,6 +6,7 @@ from datetime import datetime from typing import Callable from unittest import mock from gn3.commands import compose_gemma_cmd +from gn3.commands import compose_rqtl_cmd from gn3.commands import queue_cmd from gn3.commands import run_cmd from gn3.exceptions import RedisConnectionError @@ -53,6 +54,31 @@ class TestCommands(unittest.TestCase): "-p /tmp/gf13Ad0tRX/phenofile.txt" " -gk")) + def test_compose_rqtl_cmd(self): + """Test that the R/qtl cmd is composed correctly""" + self.assertEqual( + compose_rqtl_cmd(rqtl_wrapper_cmd="rqtl-wrapper", + rqtl_wrapper_kwargs={ + "g": "genofile", + "p": "phenofile", + "model": "normal", + "method": "hk", + "nperm": 1000, + "scale": "Mb", + "control": "rs123456" + }, + rqtl_wrapper_bool_kwargs=[ + "addcovar", + "interval" + ]), + ("rqtl-wrapper " + "--g genofile --p phenofile " + "--model normal --method hk " + "--nperm 1000 --scale Mb " + "--control rs123456 " + "--addcovar --interval") + ) + def test_queue_cmd_exception_raised_when_redis_is_down(self): """Test that the correct error is raised when Redis is unavailable""" self.assertRaises(RedisConnectionError, |