From 9eab14f7f6c40c401c29d11d69ac54303415ab87 Mon Sep 17 00:00:00 2001
From: Frederick Muriuki Muriithi
Date: Thu, 4 Jan 2024 11:50:04 +0300
Subject: Process sex and cross information data in "covar" files.

---
 r_qtl/r_qtl2.py                                  |  31 +++++++++++++++++++++++
 tests/r_qtl/test_files/test_covar.zip            | Bin 490 -> 562 bytes
 tests/r_qtl/test_files/test_covar_transposed.zip | Bin 541 -> 594 bytes
 tests/r_qtl/test_r_qtl2_covar.py                 |  30 ++++++++++++----------
 4 files changed, 48 insertions(+), 13 deletions(-)

diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py
index 26e8c76..682542b 100644
--- a/r_qtl/r_qtl2.py
+++ b/r_qtl/r_qtl2.py
@@ -134,6 +134,37 @@ def make_process_data_geno(cdata) -> tuple[
             for items in zip(ids, vals[1:]))
     return (__non_transposed__, __transposed__)
 
+def make_process_data_covar(cdata) -> tuple[
+        Callable[[dict], dict],
+        Callable[[str, tuple[str, ...], tuple[str, ...]],
+                 tuple[dict, ...]]]:
+    """Build functions to process sex and cross information in covar files."""
+    def replace_sex_code(val):
+        sex_info = cdata.get("sex", False)
+        if bool(sex_info):
+            return sex_info.get(val, val)
+        return val
+    def replace_cross_info(val):
+        cross_info = cdata.get("cross_info", False)
+        if bool(cross_info):
+            return cross_info.get(val, val)
+        return val
+    def non_transposed(row: dict) -> dict:
+        return {
+            key: thread_op(value, replace_sex_code, replace_cross_info)
+            for key,value in row.items()
+        }
+    def transposed(id_key: str,
+                   ids: tuple[str, ...],
+                   vals: tuple[str, ...]) -> tuple[dict, ...]:
+        return tuple(
+            dict(zip(
+                [id_key, vals[0]],
+                (thread_op(item, replace_sex_code, replace_cross_info)
+                 for item in items)))
+            for items in zip(ids, vals[1:]))
+    return (non_transposed, transposed)
+
 def __default_process_value_transposed__(
         id_key: str,
         ids: tuple[str, ...],
diff --git a/tests/r_qtl/test_files/test_covar.zip b/tests/r_qtl/test_files/test_covar.zip
index 6fbb5e9..61961a0 100644
Binary files a/tests/r_qtl/test_files/test_covar.zip and b/tests/r_qtl/test_files/test_covar.zip differ
diff --git a/tests/r_qtl/test_files/test_covar_transposed.zip b/tests/r_qtl/test_files/test_covar_transposed.zip
index 5e933cc..c0758a3 100644
Binary files a/tests/r_qtl/test_files/test_covar_transposed.zip and b/tests/r_qtl/test_files/test_covar_transposed.zip differ
diff --git a/tests/r_qtl/test_r_qtl2_covar.py b/tests/r_qtl/test_r_qtl2_covar.py
index c18ee74..eef7a33 100644
--- a/tests/r_qtl/test_r_qtl2_covar.py
+++ b/tests/r_qtl/test_r_qtl2_covar.py
@@ -10,19 +10,21 @@ from r_qtl import r_qtl2 as rqtl2
 @pytest.mark.parametrize(
     "filepath,expected",
     (("tests/r_qtl/test_files/test_covar.zip",
-      ({"id": "1", "sex": "m", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "2", "sex": "m", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "3", "sex": "m", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "146", "sex": "f", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "147", "sex": "f", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "148", "sex": "f", "cross_direction": "(BxS)x(BxS)"})),
+      ({"id": "1", "sex": "male", "cross_direction": 1},
+       {"id": "2", "sex": "male", "cross_direction": 1},
+       {"id": "3", "sex": "male", "cross_direction": 1},
+       {"id": "71", "sex": "male", "cross_direction": 0},
+       {"id": "72", "sex": "male", "cross_direction": 0},
+       {"id": "146", "sex": "female", "cross_direction": 1},
+       {"id": "147", "sex": "female", "cross_direction": 1},
+       {"id": "148", "sex": "female", "cross_direction": 1})),
      ("tests/r_qtl/test_files/test_covar_transposed.zip",
-      ({"id": "1", "sex": "m", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "2", "sex": "m", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "3", "sex": "m", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "146", "sex": "f", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "147", "sex": "f", "cross_direction": "(BxS)x(BxS)"},
-       {"id": "148", "sex": "f", "cross_direction": "(BxS)x(BxS)"})),
+      ({"id": "1", "sex": "male", "cross_direction": 1},
+       {"id": "2", "sex": "male", "cross_direction": 1},
+       {"id": "3", "sex": "male", "cross_direction": 1},
+       {"id": "146", "sex": "female", "cross_direction": 1},
+       {"id": "147", "sex": "female", "cross_direction": 1},
+       {"id": "148", "sex": "female", "cross_direction": 1})),
      ))
 def test_parse_covar_files(filepath, expected):
     """Test parsing of 'covar' files from the R/qtl2 bundle.
@@ -33,4 +35,6 @@ def test_parse_covar_files(filepath, expected):
     """
     with ZipFile(Path(filepath).absolute(), "r") as zfile:
         cdata = rqtl2.control_data(zfile)
-        assert tuple(rqtl2.file_data(zfile, "covar", cdata)) == expected
+        process_fns = rqtl2.make_process_data_covar(cdata)
+        assert tuple(
+            rqtl2.file_data(zfile, "covar", cdata, *process_fns)) == expected
-- 
cgit v1.2.3