Merge pull request #377 from sanger/DPL-018-dont-sequence-ZZA-controls

DPL-018 Don't sequence control sample with root sample ID prefix “ZZA” [Heron]

.release-version

@@ -1 +1 @@
-1.18.0
+1.19.0

README.md

@@ -204,7 +204,7 @@ A sample is filtered positive if:
 More information on this version can be found on [this](https://ssg-confluence.internal.sanger.ac.uk/display/PSDPUB/UAT+6th+October+2020)
 Confluence page.
 
-#### Version 2 `v2` - **Current Version**
+#### Version 2 `v2`
 
 A sample is filtered positive if:
 
@@ -215,6 +215,14 @@ A sample is filtered positive if:
 More information on this version can be found on [this](https://ssg-confluence.internal.sanger.ac.uk/display/PSDPUB/Fit+to+pick+-+v2)
 Confluence page.
 
+#### Version 3 `v3` - **Current Version**
+
+A sample is filtered positive if:
+
+- it has a 'Positive' RESULT
+- it is not a control (ROOT_SAMPLE_ID does not start with 'CBIQA_', 'QC0', or 'ZZA')
+- all of CH1_CQ, CH2_CQ and CH3_CQ are `None`, or one of these is less than or equal to 30
+
 #### Propagating Filtered Positive version changes to MongoDB, MLWH and (optionally) DART
 
 On changing the positive filtering version/definition, all unpicked samples stored in MongoDB, MLWH and DART need

crawler/filtered_positive_identifier.py

@@ -19,6 +19,7 @@
 FILTERED_POSITIVE_VERSION_0 = "v0"  # pre-filtered_positive definitions
 FILTERED_POSITIVE_VERSION_1 = "v1"  # initial implementation, as per GPL-669
 FILTERED_POSITIVE_VERSION_2 = "v2"  # updated as per GPL-699 and GPL-740
+FILTERED_POSITIVE_VERSION_3 = "v3"  # updated as per DPL-018
 
 
 class FilteredPositiveIdentifier(ABC):
@@ -81,7 +82,7 @@ def current_filtered_positive_identifier() -> FilteredPositiveIdentifier:
     Returns:
         {FilteredPositiveIdentifier} -- the current filtered positive identifier
     """
-    return FilteredPositiveIdentifierV2()
+    return FilteredPositiveIdentifierV3()
 
 
 def filtered_positive_identifier_by_version(version: str) -> FilteredPositiveIdentifier:
@@ -99,6 +100,8 @@ def filtered_positive_identifier_by_version(version: str) -> FilteredPositiveIde
         return FilteredPositiveIdentifierV1()
     elif version == FILTERED_POSITIVE_VERSION_2:
         return FilteredPositiveIdentifierV2()
+    elif version == FILTERED_POSITIVE_VERSION_3:
+        return FilteredPositiveIdentifierV3()
     else:
         raise ValueError(f"'{version}' is not a known filtered positive version")
 
@@ -123,3 +126,11 @@ def __init__(self):
         self.version = FILTERED_POSITIVE_VERSION_2
         self.root_sample_id_control_regex = re.compile("^(?:CBIQA_|QC0|ZZA000)")
         self.evaluate_ct_values = True
+
+
+class FilteredPositiveIdentifierV3(FilteredPositiveIdentifier):
+    def __init__(self):
+        super(FilteredPositiveIdentifierV3, self).__init__()
+        self.version = FILTERED_POSITIVE_VERSION_3
+        self.root_sample_id_control_regex = re.compile("^(?:CBIQA_|QC0|ZZA)")
+        self.evaluate_ct_values = True

tests/test_filtered_positive_identifier.py

@@ -14,9 +14,11 @@
     FILTERED_POSITIVE_VERSION_0,
     FILTERED_POSITIVE_VERSION_1,
     FILTERED_POSITIVE_VERSION_2,
+    FILTERED_POSITIVE_VERSION_3,
     FilteredPositiveIdentifierV0,
     FilteredPositiveIdentifierV1,
     FilteredPositiveIdentifierV2,
+    FilteredPositiveIdentifierV3,
     current_filtered_positive_identifier,
     filtered_positive_identifier_by_version,
 )
@@ -39,7 +41,7 @@ def positive_sample():
 
 def test_current_filtered_positive_identifier():
     identifier = current_filtered_positive_identifier()
-    assert identifier.version == FILTERED_POSITIVE_VERSION_2
+    assert identifier.version == FILTERED_POSITIVE_VERSION_3
 
 
 # ----- tests for filtered_positive_identifier_by_version() -----
@@ -60,6 +62,11 @@ def test_filtered_positive_identifier_by_version_returns_V2_identifier():
     assert identifier.version == FILTERED_POSITIVE_VERSION_2
 
 
+def test_filtered_positive_identifier_by_version_returns_V3_identifier():
+    identifier = filtered_positive_identifier_by_version(FILTERED_POSITIVE_VERSION_3)
+    assert identifier.version == FILTERED_POSITIVE_VERSION_3
+
+
 def test_filtered_positive_identifier_by_version_raises_unknown_version():
     with pytest.raises(ValueError):
         filtered_positive_identifier_by_version("unknown version")
@@ -301,3 +308,110 @@ def test_v2_is_positive_returns_false_all_ct_values_greater_than_30():
     sample[FIELD_CH2_CQ] = Decimal128("41.12345678")
     sample[FIELD_CH3_CQ] = Decimal128("42.12345678")
     assert identifier.is_positive(sample) is False
+
+
+# ----- tests for FilteredPositiveIdentifierV3 -----
+
+
+def test_v3_version():
+    identifier = FilteredPositiveIdentifierV3()
+    assert identifier.version == FILTERED_POSITIVE_VERSION_3
+
+
+def test_v3_is_positive_returns_true_matching_criteria():
+    identifier = FilteredPositiveIdentifierV3()
+
+    # expected positive match
+    sample = positive_sample()
+    assert identifier.is_positive(sample) is True
+
+    # case invariant positive match
+    sample = positive_sample()
+    sample[FIELD_RESULT] = "POSITIVE"
+    assert identifier.is_positive(sample) is True
+
+    # 3x mix of ct values
+    sample = positive_sample()
+    sample[FIELD_CH2_CQ] = Decimal128("41.12345678")
+    sample[FIELD_CH3_CQ] = None
+    assert identifier.is_positive(sample) is True
+
+    sample = positive_sample()
+    sample[FIELD_CH1_CQ] = None
+    sample[FIELD_CH3_CQ] = Decimal128("42.12345678")
+    assert identifier.is_positive(sample) is True
+
+    sample = positive_sample()
+    sample[FIELD_CH1_CQ] = Decimal128("40.12345678")
+    sample[FIELD_CH2_CQ] = None
+    assert identifier.is_positive(sample) is True
+
+    # all ct values None
+    sample = positive_sample()
+    sample[FIELD_CH1_CQ] = None
+    sample[FIELD_CH2_CQ] = None
+    sample[FIELD_CH3_CQ] = None
+    assert identifier.is_positive(sample) is True
+
+    # no FIELD_CHX_CQ fields
+    sample = {FIELD_RESULT: RESULT_VALUE_POSITIVE, FIELD_ROOT_SAMPLE_ID: "MCM001"}
+    assert identifier.is_positive(sample) is True
+
+
+def test_v3_is_positive_returns_false_result_not_positive():
+    identifier = FilteredPositiveIdentifierV3()
+
+    # does not conform to regex
+    sample = positive_sample()
+    sample[FIELD_RESULT] = "  positive"
+    assert identifier.is_positive(sample) is False
+
+    # negative result
+    sample = positive_sample()
+    sample[FIELD_RESULT] = "Negative"
+    assert identifier.is_positive(sample) is False
+
+    # void result
+    sample = positive_sample()
+    sample[FIELD_RESULT] = "Void"
+    assert identifier.is_positive(sample) is False
+
+    # 'limit of detection' result
+    sample = positive_sample()
+    sample[FIELD_RESULT] = RESULT_VALUE_LIMIT_OF_DETECTION
+    assert identifier.is_positive(sample) is False
+
+    # case invariant 'limit of detection' result
+    sample = positive_sample()
+    sample[FIELD_RESULT] = RESULT_VALUE_LIMIT_OF_DETECTION.upper()
+    assert identifier.is_positive(sample) is False
+
+
+def test_v3_is_positive_returns_false_control_sample():
+    identifier = FilteredPositiveIdentifierV3()
+
+    sample = positive_sample()
+    sample[FIELD_ROOT_SAMPLE_ID] = "CBIQA_MCM001"
+    assert identifier.is_positive(sample) is False
+
+    sample = positive_sample()
+    sample[FIELD_ROOT_SAMPLE_ID] = "QC0_MCM001"
+    assert identifier.is_positive(sample) is False
+
+    sample = positive_sample()
+    sample[FIELD_ROOT_SAMPLE_ID] = "ZZA000_MCM001"
+    assert identifier.is_positive(sample) is False
+
+    sample = positive_sample()
+    sample[FIELD_ROOT_SAMPLE_ID] = "ZZA3c4j_MCM001"
+    assert identifier.is_positive(sample) is False
+
+
+def test_v3_is_positive_returns_false_all_ct_values_greater_than_30():
+    identifier = FilteredPositiveIdentifierV3()
+
+    sample = positive_sample()
+    sample[FIELD_CH1_CQ] = Decimal128("40.12345678")
+    sample[FIELD_CH2_CQ] = Decimal128("41.12345678")
+    sample[FIELD_CH3_CQ] = Decimal128("42.12345678")
+    assert identifier.is_positive(sample) is False