From 4bf51a31c251f3d788325c58c8a78f534409f90f Mon Sep 17 00:00:00 2001
From: Frederick Muriuki Muriithi
Date: Wed, 3 Jan 2024 10:10:12 +0300
Subject: Parse the phenotype data from the R/qtl2 bundle.

---
 r_qtl/r_qtl2.py                                  |  13 +++++++++
 tests/r_qtl/test_files/test_pheno.zip            | Bin 0 -> 485 bytes
 tests/r_qtl/test_files/test_pheno_transposed.zip | Bin 0 -> 536 bytes
 tests/r_qtl/test_r_qtl2_pheno.py                 |  33 +++++++++++++++++++++++
 4 files changed, 46 insertions(+)
 create mode 100644 tests/r_qtl/test_files/test_pheno.zip
 create mode 100644 tests/r_qtl/test_files/test_pheno_transposed.zip
 create mode 100644 tests/r_qtl/test_r_qtl2_pheno.py

diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py
index bc3a86f..2256609 100644
--- a/r_qtl/r_qtl2.py
+++ b/r_qtl/r_qtl2.py
@@ -163,3 +163,16 @@ def map_data(zfile: ZipFile, map_type: str, cdata: dict) -> Iterator[dict]:
 
     for row in with_transposed(zfile, map_file_key, cdata, __merge__):
         yield row
+
+def phenotype_data(zfile: ZipFile, cdata: dict) -> Iterator[dict]:
+    """Load phenotype file data."""
+    if not cdata.get("pheno_transposed", False):
+        for row in with_non_transposed(zfile, "pheno", cdata, lambda val: val):
+            yield row
+        return
+
+    def __merge__(id_key, ids, vals):
+        return tuple(dict(zip([id_key, vals[0]], items))
+                     for items in zip(ids, vals[1:]))
+    for row in with_transposed(zfile, "pheno", cdata, __merge__):
+        yield row
diff --git a/tests/r_qtl/test_files/test_pheno.zip b/tests/r_qtl/test_files/test_pheno.zip
new file mode 100644
index 0000000..5c709e7
Binary files /dev/null and b/tests/r_qtl/test_files/test_pheno.zip differ
diff --git a/tests/r_qtl/test_files/test_pheno_transposed.zip b/tests/r_qtl/test_files/test_pheno_transposed.zip
new file mode 100644
index 0000000..9bff030
Binary files /dev/null and b/tests/r_qtl/test_files/test_pheno_transposed.zip differ
diff --git a/tests/r_qtl/test_r_qtl2_pheno.py b/tests/r_qtl/test_r_qtl2_pheno.py
new file mode 100644
index 0000000..554d9c8
--- /dev/null
+++ b/tests/r_qtl/test_r_qtl2_pheno.py
@@ -0,0 +1,33 @@
+"""Test parsing of pheno files in a R/qtl2 bundle."""
+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(
+    "filepath,expected",
+    (("tests/r_qtl/test_files/test_pheno.zip",
+      ({"id": "1", "liver": "61.92", "spleen": "153.16"},
+       {"id": "2", "liver": "88.33", "spleen": "178.58"},
+       {"id": "3", "liver": "58", "spleen": "131.91"},
+       {"id": "4", "liver": "78.06", "spleen": "126.13"},
+       {"id": "5", "liver": "65.31", "spleen": "181.05"})),
+     ("tests/r_qtl/test_files/test_pheno_transposed.zip",
+      ({"id": "1", "liver": "61.92", "spleen": "153.16"},
+       {"id": "2", "liver": "88.33", "spleen": "178.58"},
+       {"id": "3", "liver": "58", "spleen": "131.91"},
+       {"id": "4", "liver": "78.06", "spleen": "126.13"},
+       {"id": "5", "liver": "65.31", "spleen": "181.05"}))))
+def test_parse_pheno_files(filepath, expected):
+    """Test parsing of 'pheno' files from the R/qtl2 bundle.
+
+    GIVEN: path to a R/qtl2 bundle
+    WHEN: the 'pheno' file is parsed
+    THEN: verify the parsed data is as expected
+    """
+    with ZipFile(Path(filepath).absolute(), "r") as zfile:
+        assert tuple(
+            rqtl2.phenotype_data(zfile, rqtl2.control_data(zfile))) == expected
-- 
cgit v1.2.3