From 46410163ca402660434b09a8fc8ec7b12b89d8a1 Mon Sep 17 00:00:00 2001
From: Frederick Muriuki Muriithi
Date: Wed, 27 Dec 2023 08:08:31 +0300
Subject: Tests: Add tests for parsing gmap files.

---
 pytest.ini                                      |   3 +-
 r_qtl/r_qtl2.py                                 |  31 +++++++++-------
 tests/r_qtl/test_files/test_gmap.zip            | Bin 0 -> 542 bytes
 tests/r_qtl/test_files/test_gmap_transposed.zip | Bin 0 -> 594 bytes
 tests/r_qtl/test_r_qtl2.py                      |  47 ++++++++++++++++++++++++
 5 files changed, 67 insertions(+), 14 deletions(-)
 create mode 100644 tests/r_qtl/test_files/test_gmap.zip
 create mode 100644 tests/r_qtl/test_files/test_gmap_transposed.zip
 create mode 100644 tests/r_qtl/test_r_qtl2.py

diff --git a/pytest.ini b/pytest.ini
index 9fd309c..99e2c59 100644
--- a/pytest.ini
+++ b/pytest.ini
@@ -1,4 +1,5 @@
 [pytest]
 addopts = --strict-markers
 markers =
-	slow
\ No newline at end of file
+	slow
+	unit_test
\ No newline at end of file
diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py
index 26d6dd4..a221f26 100644
--- a/r_qtl/r_qtl2.py
+++ b/r_qtl/r_qtl2.py
@@ -65,20 +65,27 @@ def genotype_data(zfile: ZipFile, cdata: dict) -> Iterator[dict]:
 def map_data(zfile: ZipFile, map_type: str, cdata: dict) -> dict:
     """Read gmap files to get the genome mapping data"""
     assert map_type in ("genetic-map", "physical-map"), "Invalid map type"
-    map_key = {"genetic-map": "gmap", "physical-map": "pmap"}[map_type]
-    if not cdata.get("geno_transposed", False):
-        with zfile.open(cdata[map_key]) as gmapfile:
+    map_file = cdata[{
+        "genetic-map": "gmap",
+        "physical-map": "pmap"
+    }[map_type]]
+    # TODO: Might need to check `gmap_transposed` and `pmap_transposed` instead
+    #  of `geno_transposed` -- see
+    #  https://github.com/rqtl/qtl2data/blob/main/ArabMAGIC/arabmagic_tair8.json
+    #  for the *_transposed values
+    transposed_dict = {
+        "genetic-map": "gmap_transposed",
+        "physical-map": "pmap_transposed"
+    }
+    if not cdata.get(transposed_dict[map_type], False):
+        with zfile.open(map_file) as gmapfile:
             reader = csv.DictReader(
                 filter(lambda line: not line.startswith("#"),
                        io.TextIOWrapper(gmapfile)),
                 delimiter=cdata.get("sep", ","))
-            return {
-                line["marker"]: {
-                    key: value for key,value in line.items() if key != "marker"
-                } for line in reader
-            }
+            return tuple(row for row in reader)
 
-    with zfile.open(cdata["gmap"]) as gmapfile:
+    with zfile.open(map_file) as gmapfile:
         lines = [[field.strip() for field in
                   line.strip().split(cdata.get("sep", ","))]
                  for line in
@@ -87,11 +94,9 @@ def map_data(zfile: ZipFile, map_type: str, cdata: dict) -> dict:
 
     headers = tuple(line[0] for line in lines)
     return reduce(
-        lambda gmap, row: {
-            **gmap,
-            row[0]: dict(zip(headers[1:], row[1:]))},
+        lambda gmap, row: gmap + (dict(zip(headers, row)),),
         zip(*(line[1:] for line in lines)),
-        {})
+        tuple())
 
 def read_r_qtl2_files(filepath: Path):
     """Read R/qtl2 format zip files."""
diff --git a/tests/r_qtl/test_files/test_gmap.zip b/tests/r_qtl/test_files/test_gmap.zip
new file mode 100644
index 0000000..c8d452a
Binary files /dev/null and b/tests/r_qtl/test_files/test_gmap.zip differ
diff --git a/tests/r_qtl/test_files/test_gmap_transposed.zip b/tests/r_qtl/test_files/test_gmap_transposed.zip
new file mode 100644
index 0000000..74a956e
Binary files /dev/null and b/tests/r_qtl/test_files/test_gmap_transposed.zip differ
diff --git a/tests/r_qtl/test_r_qtl2.py b/tests/r_qtl/test_r_qtl2.py
new file mode 100644
index 0000000..a72b300
--- /dev/null
+++ b/tests/r_qtl/test_r_qtl2.py
@@ -0,0 +1,47 @@
+"""Test the parsing of the R/qtl2 data 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_gmap.zip",
+      ({"marker": "PVV4", "chr": "1", "pos": "0.000000"},
+       {"marker": "AXR-1", "chr": "1", "pos": "6.250674"},
+       {"marker": "HH.335C-Col/PhyA", "chr": "1", "pos": "9.303868"},
+       {"marker": "EC.480C", "chr": "1", "pos": "12.577629"},
+       {"marker": "EC.66C", "chr": "1", "pos": "18.392830"})),
+     ("tests/r_qtl/test_files/test_gmap_transposed.zip",
+      ({"marker": "PVV4", "chr": "1", "pos": "0.000000"},
+       {"marker": "AXR-1", "chr": "1", "pos": "6.250674"},
+       {"marker": "HH.335C-Col/PhyA", "chr": "1", "pos": "9.303868"},
+       {"marker": "EC.480C", "chr": "1", "pos": "12.577629"},
+       {"marker": "EC.66C", "chr": "1", "pos": "18.392830"}))))
+def test_parse_gmap_files(relpath, expected):
+    """
+    GIVEN: A path to a zip file, `relpath`
+    WHEN: we parse a gmap file
+    THEN: ensure the parsed data is as expected.
+    """
+    with ZipFile(Path(relpath).absolute(), "r") as zfile:
+        assert rqtl2.map_data(
+            zfile, "genetic-map", rqtl2.control_data(zfile)) == expected
+
+@pytest.mark.skip(reason="Test not implemented yet.")
+@pytest.mark.unit_test
+@pytest.mark.parametrize(
+    "relpath,expected",
+    (("tests/r_qtl/test_files/test_pmap.zip",
+      ()),
+     ("tests/r_qtl/test_files/test_pmap_transposed.zip",
+      ())))
+def test_parse_pmap_files(relpath, expected):
+    """
+    GIVEN: A path to a zip file, `relpath`
+    WHEN: we parse a pmap file
+    THEN: ensure the parsed data is as expected.
+    """
-- 
cgit v1.2.3