"""Test the parsing of the R/qtl2 geno files."""
from pathlib import Path
import pytest
from zipfile import ZipFile
from r_qtl import r_qtl2 as rqtl2
@pytest.mark.unit_test
@pytest.mark.parametrize(
"relpath,expected",
(
("tests/r_qtl/test_files/test_geno.zip",
({
"id": "1",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
},
{
"id": "2",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
},
{
"id": "3",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": None,
"EC.480C": 1,
"EC.66C": 1
},
{
"id": "4",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
},
{
"id": "5",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
},
{
"id": "6",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
},
{
"id": "7",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
},
{
"id": "8",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 1,
"EC.66C": 1
},
{
"id": "9",
"PVV4": None,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
},
{
"id": "10",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
})),
("tests/r_qtl/test_files/test_geno_transposed.zip",
({
"id": "1",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "2",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "3",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": None,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "4",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
},{
"id": "5",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
}, {
"id": "6",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
}, {
"id": "7",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "8",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "9",
"PVV4": None,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
}, {
"id": "10",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
}))))
def test_parse_geno_files(relpath, expected):
"""Test parsing of geno files from the R/qtl2 bundle.
GIVEN: Path to a zip file with R/qtl2 data
WHEN: we parse the geno file
THEN: ensure that the data we get is as expected
"""
with ZipFile(Path(relpath).absolute(), "r") as zfile:
cdata = rqtl2.control_data(zfile)
assert tuple(rqtl2.file_data(
zfile,
"geno",
cdata,
*rqtl2.make_process_data_geno(cdata))) == expected
@pytest.mark.unit_test
@pytest.mark.parametrize(
"relpath,expected",
(("tests/r_qtl/test_files/test_founder_geno.zip",
(({
"id": "1",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "2",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "3",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": None,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "4",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "5",
"PVV4": None,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
}))),
("tests/r_qtl/test_files/test_founder_geno_transposed.zip",
(({
"id": "1",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "2",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "3",
"PVV4": 2,
"AXR-1": 2,
"HH.335C-Col/PhyA": None,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "4",
"PVV4": 1,
"AXR-1": 1,
"HH.335C-Col/PhyA": 1,
"EC.480C": 1,
"EC.66C": 1
}, {
"id": "5",
"PVV4": None,
"AXR-1": 2,
"HH.335C-Col/PhyA": 2,
"EC.480C": 2,
"EC.66C": 2
})))))
def test_parse_founder_geno_files(relpath, expected):
"""Test parsing of founder_geno files from the R/qtl2 bundle.
GIVEN: Path to a zip file with R/qtl2 data
WHEN: we parse the founder_geno file
THEN: ensure that the data we get is as expected
"""
with ZipFile(Path(relpath).absolute(), "r") as zfile:
cdata = rqtl2.control_data(zfile)
assert tuple(rqtl2.file_data(
zfile,
"founder_geno",
cdata,
*rqtl2.make_process_data_geno(cdata))) == expected
@pytest.mark.unit_test
@pytest.mark.parametrize(
"filepath,member,transposed,expected",
(("tests/r_qtl/test_files/test_geno.zip",
"test_geno.csv",
False,
("1", "2", "3", "4", "5", "6", "7", "8", "9", "10")),
("tests/r_qtl/test_files/test_geno_transposed.zip",
"test_geno_transposed.csv",
True,
("1", "2", "3", "4", "5", "6", "7", "8", "9", "10")),
("tests/r_qtl/test_files/test_geno_multiple.zip",
"test_geno_multiple01.csv",
False,
("1", "2", "3", "4", "5")),
("tests/r_qtl/test_files/test_geno_multiple.zip",
"test_geno_multiple02.csv",
False,
("6", "7", "8", "9", "10"))))
def test_load_geno_samples(filepath, member, transposed, expected):
assert sorted(rqtl2.load_samples(
filepath, member, transposed)) == sorted(expected)
