From 3f879d120f7628646f383a457206b15037cc57dc Mon Sep 17 00:00:00 2001
From: Frederick Muriuki Muriithi
Date: Mon, 5 Feb 2024 15:18:44 +0300
Subject: Check that data in geno file is valid

Add a function to ensure the values in the geno files are all listed
in the control data under the "genotypes" key.
---
 r_qtl/r_qtl2.py    |  4 ++--
 r_qtl/r_qtl2_qc.py | 18 +++++++++++++++++-
 2 files changed, 19 insertions(+), 3 deletions(-)

(limited to 'r_qtl')

diff --git a/r_qtl/r_qtl2.py b/r_qtl/r_qtl2.py
index f8c08d9..e74312f 100644
--- a/r_qtl/r_qtl2.py
+++ b/r_qtl/r_qtl2.py
@@ -10,7 +10,7 @@ import yaml
 
 from functional_tools import take, chain
 
-from r_qtl.errors import InvalidFormat
+from r_qtl.errors import InvalidFormat, MissingFileError
 
 __FILE_TYPES__ = (
     "geno", "founder_geno", "pheno", "covar", "phenocovar", "gmap", "pmap",
@@ -250,7 +250,7 @@ def file_data(zfile: ZipFile,
                 zfile, member_key, cdata, process_transposed_value):
             yield row
     except KeyError as exc:
-        raise InvalidFormat(*exc.args) from exc
+        raise MissingFileError(*exc.args) from exc
 
 def cross_information(zfile: ZipFile, cdata: dict) -> Iterator[dict]:
     """Load cross information where present."""
diff --git a/r_qtl/r_qtl2_qc.py b/r_qtl/r_qtl2_qc.py
index 853c34b..70a00f9 100644
--- a/r_qtl/r_qtl2_qc.py
+++ b/r_qtl/r_qtl2_qc.py
@@ -1,7 +1,7 @@
 """Quality control checks for R/qtl2 data bundles."""
 from zipfile import ZipFile
 from functools import reduce
-from typing import Union, Sequence
+from typing import Union, Sequence, Iterator, Optional
 
 from r_qtl import errors as rqe
 from r_qtl import r_qtl2 as rqtl2
@@ -54,3 +54,19 @@ def validate_bundle(zfile: ZipFile):
         raise rqe.MissingFileError(
                         "The following files do not exist in the bundle: " +
                         ", ".join(missing))
+
+def geno_errors(zfile: ZipFile) -> Iterator[tuple[Optional[int], Optional[str], str]]:
+    """Check for and retrieve geno errors."""
+    cdata = rqtl2.control_data(zfile)
+    genotypes = tuple(cdata.get("genotypes", {}).keys())
+    try:
+        for lineno, row in enumerate(
+                rqtl2.file_data(zfile, "geno", cdata), start=1):
+            for field, value in row.items():
+                if field == "id":
+                    continue
+                if value is not None and value not in genotypes:
+                    yield (lineno, field, (
+                        f"Invalid value '{value}'. Expected one of {genotypes}"))
+    except rqe.MissingFileError:
+        yield (None, None, "Missing 'geno' file.")
-- 
cgit v1.2.3