Merge pull request #112 from eastgenomics/improve_url_value_typing

Fix url value typing

dxapp.json

@@ -3,15 +3,15 @@
   "title": "eggd_generate_variant_workbook",
   "summary": "Create Excel workbook from VEP annotated vcf",
   "dxapi": "1.0.0",
-  "version": "2.1.0",
+  "version": "2.1.1",
   "authorizedUsers": [
     "org-emee_1"
   ],
   "developers":[
     "org-emee_1"
   ],
   "properties": {
-    "githubRelease": "v2.1.0"
+    "githubRelease": "v2.1.1"
   },
   "inputSpec": [
     {

resources/home/dnanexus/generate_workbook/tests/test_utils.py

@@ -0,0 +1,25 @@
+import os
+import sys
+
+sys.path.append(os.path.abspath(
+    os.path.join(os.path.realpath(__file__), '../../')
+))
+
+from utils.utils import is_numeric
+
+
+def test_is_numeric():
+    """
+    Tests that all forms of numerical value are correctly identified as
+    being numeric
+    """
+    numeric_values = ['5', '5.5', '-0.1', '1e-5']
+    non_numeric_values = ['5,6', '5, 6', '.', '3/4', ]
+
+    assert all([is_numeric(x) for x in numeric_values]), (
+        'some numeric values do not properly evaluate to being numeric'
+    )
+
+    assert all([not is_numeric(x) for x in non_numeric_values]), (
+        'some non-numeric values wrongly evaluate to being numeric'
+    )

resources/home/dnanexus/generate_workbook/utils/excel.py

@@ -10,6 +10,7 @@
 from openpyxl.styles.fills import PatternFill
 import pandas as pd
 
+from .utils import is_numeric
 
 # openpyxl style settings
 THIN = Side(border_style="thin", color="000000")
@@ -531,28 +532,10 @@ def set_types(self, worksheet) -> None:
         """
         for cells in worksheet.rows:
             for cell in cells:
-                if self.is_numeric(cell.value):
+                if is_numeric(cell.value):
                     cell.data_type = 'n'
 
 
-    def is_numeric(self, value) -> bool:
-        """
-        Returns true if given value is in some form numeric
-
-        Parameters
-        ----------
-        value : str
-            string to check if can be cast to numeric type
-
-        Returns
-        -------
-        bool
-            True if value can be numeric
-        """
-        return str(value).lstrip('-').replace('.', '').replace(
-            'e-', '', 1).replace('e', '').isdigit()
-
-
     def set_font(self, worksheet) -> None:
         """
         Set font to all cells in variant sheet to Calibri

resources/home/dnanexus/generate_workbook/utils/utils.py

@@ -0,0 +1,16 @@
+def is_numeric(value:str) -> bool:
+    """
+    Returns true if given value is in some form numeric
+
+    Parameters
+    ----------
+    value : str
+        string to check if can be cast to numeric type
+
+    Returns
+    -------
+    bool
+        True if value can be numeric
+    """
+    return str(value).lstrip('-').replace('.', '').replace(
+        'e-', '', 1).replace('e', '').isdigit()

resources/home/dnanexus/generate_workbook/utils/vcf.py

@@ -10,6 +10,7 @@
 
 from .columns import splitColumns
 from .filters import filter
+from .utils import is_numeric
 
 
 class vcf():
@@ -483,7 +484,7 @@ def make_hyperlink(self, column, url, value, build):
             # If URL is too long just display the value
             return value[column]
 
-        if any([value[column].endswith(x) for x in ['_AC', '_AF', '_AN']]):
+        if is_numeric(value[column]):
             # return numeric values not wrapped in quotes
             return f'=HYPERLINK("{url}", {value[column]})'
         else: