From 7b94f989bcfbf6543bfa628422331adfa3d5daac Mon Sep 17 00:00:00 2001
From: BonfaceKilz
Date: Wed, 24 Mar 2021 16:18:00 +0300
Subject: Add extra procedure for parsing a genotype file

* gn3/computations/parsers.py (parse_genofile): New procedure.
* tests/unit/computations/test_parsers.py: New test files for above.
---
 gn3/computations/parsers.py | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100644 gn3/computations/parsers.py

(limited to 'gn3/computations/parsers.py')

diff --git a/gn3/computations/parsers.py b/gn3/computations/parsers.py
new file mode 100644
index 0000000..94387ff
--- /dev/null
+++ b/gn3/computations/parsers.py
@@ -0,0 +1,38 @@
+"""Parsers for generating some files in genenetwork"""
+import os
+from typing import Any, Dict, List, Tuple
+
+
+def parse_genofile(file_path: str) -> Tuple[List[str],
+                                            List[Dict[str, Any]]]:
+    """Parse a genotype file with a given format"""
+    if not os.path.exists(file_path):
+        raise FileNotFoundError
+    __map = {
+        'b': -1,
+        'd': 1,
+        'h': 0,
+        'u': None,
+    }
+    genotypes, strains = [], []
+    with open(file_path, "r") as _genofile:
+        for line in _genofile:
+            line = line.strip()
+            if line.startswith(("#", "@")):
+                continue
+            cells = line.split()
+            if line.startswith("Chr"):
+                strains = cells[4:]
+                strains = [strain.lower() for strain in strains]
+                continue
+            values = [__map.get(value.lower(), None) for value in cells[4:]]
+            genotype = {
+                "chr": cells[0],
+                "locus": cells[1],
+                "cm": cells[2],
+                "mb": cells[3],
+                "values":  values,
+                "dicvalues": dict(zip(strains, values)),
+            }
+            genotypes.append(genotype)
+        return strains, genotypes
-- 
cgit v1.2.3