add raredisease pedigree check

cg/constants/nf_analysis.py

@@ -18,12 +18,16 @@ class NfTowerStatus(StrEnum):
 RAREDISEASE_METRIC_CONDITIONS: dict[str, dict[str, Any]] = {
     "percent_duplicates": {"norm": "lt", "threshold": 20},
     "PCT_PF_UQ_READS_ALIGNED": {"norm": "gt", "threshold": 0.95},
-    "MEDIAN_TARGET_COVERAGE": {"norm": "gt", "threshold": 26},
+    "MEDIAN_TARGET_COVERAGE": {"norm": "gt", "threshold": 25},
     "PCT_TARGET_BASES_10X": {"norm": "gt", "threshold": 0.95},
     "PCT_EXC_ADAPTER": {"norm": "lt", "threshold": 0.0005},
     "predicted_sex_sex_check": {"norm": "eq", "threshold": None},
 }
 
+RAREDISEASE_PARENT_PEDDY_METRIC_CONDITION: dict[str, dict[str, Any]] = {
+    "parent_error_ped_check": {"norm": "eq", "threshold": "False"},
+}
+
 RNAFUSION_METRIC_CONDITIONS: dict[str, dict[str, Any]] = {
     "uniquely_mapped_percent": {"norm": "gt", "threshold": 60},
     "PCT_MRNA_BASES": {"norm": "gt", "threshold": 80},

cg/meta/workflow/nf_analysis.py

@@ -631,11 +631,10 @@ def get_workflow_metrics(self, metric_id: str) -> dict:
     def get_multiqc_search_patterns(self, case_id: str) -> dict:
         """Return search patterns for MultiQC. Each key is a search pattern and each value
         corresponds to the metric ID to set in the metrics deliverables file.
-        Multiple search patterns can be added. Ideally patterns used should be sample ids, e.g.
+        Multiple search patterns can be added. Ideally, used patterns should be sample ids, e.g.
         {sample_id_1: sample_id_1, sample_id_2: sample_id_2}."""
-        sample_ids: Iterator[str] = self.status_db.get_sample_ids_by_case_id(case_id=case_id)
-        search_patterns: dict[str, str] = {sample_id: sample_id for sample_id in sample_ids}
-        return search_patterns
+        sample_ids: Iterator[str] = self.status_db.get_sample_ids_by_case_id(case_id)
+        return {sample_id: sample_id for sample_id in sample_ids}
 
     @staticmethod
     def get_deduplicated_metrics(metrics: list[MetricsBase]) -> list[MetricsBase]:
@@ -649,11 +648,13 @@ def get_deduplicated_metrics(metrics: list[MetricsBase]) -> list[MetricsBase]:
                 deduplicated_metrics.append(metric)
         return deduplicated_metrics
 
+    def get_multiqc_data_json(self, case_id: str) -> MultiqcDataJson:
+        """Return a MultiqcDataJson object."""
+        return MultiqcDataJson(**read_json(file_path=self.get_multiqc_json_path(case_id=case_id)))
+
     def get_multiqc_json_metrics(self, case_id: str) -> list[MetricsBase]:
         """Return a list of the metrics specified in a MultiQC json file."""
-        multiqc_json = MultiqcDataJson(
-            **read_json(file_path=self.get_multiqc_json_path(case_id=case_id))
-        )
+        multiqc_json: MultiqcDataJson = self.get_multiqc_data_json(case_id=case_id)
         metrics = []
         for search_pattern, metric_id in self.get_multiqc_search_patterns(case_id=case_id).items():
             metrics_for_pattern: list[MetricsBase] = (
@@ -668,6 +669,14 @@ def get_multiqc_json_metrics(self, case_id: str) -> list[MetricsBase]:
         metrics = self.get_deduplicated_metrics(metrics=metrics)
         return metrics
 
+    @staticmethod
+    def _is_pattern_found(pattern: str, text: str, exact_match: bool) -> bool:
+        if exact_match:
+            is_pattern_found: bool = pattern == text
+        else:
+            is_pattern_found: bool = pattern in text
+        return is_pattern_found
+
     def get_metrics_from_multiqc_json_with_pattern(
         self,
         search_pattern: str,
@@ -678,13 +687,11 @@ def get_metrics_from_multiqc_json_with_pattern(
         """Parse a MultiqcDataJson and returns a list of metrics."""
         metrics: list[MetricsBase] = []
         for section in multiqc_json.report_general_stats_data:
-            for section_name, section_values in section.items():
-                if exact_match:
-                    is_pattern_found: bool = search_pattern == section_name
-                else:
-                    is_pattern_found: bool = search_pattern in section_name
-                if is_pattern_found:
-                    for metric_name, metric_value in section_values.items():
+            for subsection, metrics_dict in section.items():
+                if self._is_pattern_found(
+                    pattern=search_pattern, text=subsection, exact_match=exact_match
+                ):
+                    for metric_name, metric_value in metrics_dict.items():
                         metric: MetricsBase = self.get_multiqc_metric(
                             metric_name=metric_name, metric_value=metric_value, metric_id=metric_id
                         )

cg/meta/workflow/raredisease.py

@@ -1,6 +1,7 @@
 """Module for Raredisease Analysis API."""
 
 import logging
+from itertools import permutations
 from pathlib import Path
 from typing import Any
 
@@ -20,11 +21,13 @@
     RAREDISEASE_COVERAGE_INTERVAL_TYPE,
     RAREDISEASE_COVERAGE_THRESHOLD,
     RAREDISEASE_METRIC_CONDITIONS,
+    RAREDISEASE_PARENT_PEDDY_METRIC_CONDITION,
 )
 from cg.constants.scout import RAREDISEASE_CASE_TAGS
 from cg.constants.subject import PlinkPhenotypeStatus, PlinkSex
 from cg.meta.workflow.nf_analysis import NfAnalysisAPI
 from cg.models.cg_config import CGConfig
+from cg.models.deliverables.metric_deliverables import MetricsBase, MultiqcDataJson
 from cg.models.raredisease.raredisease import (
     RarediseaseParameters,
     RarediseaseSampleSheetEntry,
@@ -149,11 +152,58 @@ def get_managed_variants(self) -> list[str]:
         return self._get_managed_variants(genome_build=GenePanelGenomeBuild.hg19)
 
     def get_workflow_metrics(self, sample_id: str) -> dict:
+        """Return Raredisease workflow metric conditions for a sample."""
         sample: Sample = self.status_db.get_sample_by_internal_id(internal_id=sample_id)
-        metric_conditions: dict[str, dict[str, Any]] = dict(RAREDISEASE_METRIC_CONDITIONS)
-        self.set_order_sex_for_sample(sample, metric_conditions)
+        if "-" not in sample_id:
+            metric_conditions: dict[str, dict[str, Any]] = RAREDISEASE_METRIC_CONDITIONS.copy()
+            self.set_order_sex_for_sample(sample, metric_conditions)
+        else:
+            metric_conditions = RAREDISEASE_PARENT_PEDDY_METRIC_CONDITION.copy()
         return metric_conditions
 
+    def _get_sample_pair_patterns(self, case_id: str) -> list[str]:
+        """Return sample-pair patterns for searching in MultiQC."""
+        sample_ids: list[str] = list(self.status_db.get_sample_ids_by_case_id(case_id=case_id))
+        pairwise_patterns: list[str] = [
+            f"{sample1}-{sample2}" for sample1, sample2 in permutations(sample_ids, 2)
+        ]
+        return pairwise_patterns
+
+    def get_parent_error_ped_check_metric(
+        self, pair_sample_ids: str, multiqc_raw_data: dict[dict]
+    ) -> MetricsBase | None:
+        """Return the parsed metrics for pedigree error given a concatenated pair of sample ids."""
+        metric_name: str = "parent_error_ped_check"
+        peddy_metrics: dict[str, dict] = multiqc_raw_data["multiqc_peddy"]
+        if sample_pair_metrics := peddy_metrics.get(pair_sample_ids, None):
+            return self.get_multiqc_metric(
+                metric_name=metric_name,
+                metric_value=sample_pair_metrics[metric_name],
+                metric_id=pair_sample_ids,
+            )
+
+    def get_multiqc_json_metrics(self, case_id: str) -> list[MetricsBase]:
+        """Return a list of the metrics specified in a MultiQC json file."""
+        multiqc_json: MultiqcDataJson = self.get_multiqc_data_json(case_id=case_id)
+        metrics = []
+        for search_pattern, metric_id in self.get_multiqc_search_patterns(case_id).items():
+            metrics_for_pattern: list[MetricsBase] = (
+                self.get_metrics_from_multiqc_json_with_pattern(
+                    search_pattern=search_pattern,
+                    multiqc_json=multiqc_json,
+                    metric_id=metric_id,
+                    exact_match=self.is_multiqc_pattern_search_exact,
+                )
+            )
+            metrics.extend(metrics_for_pattern)
+        for sample_pair in self._get_sample_pair_patterns(case_id):
+            if parent_error_metric := self.get_parent_error_ped_check_metric(
+                pair_sample_ids=sample_pair, multiqc_raw_data=multiqc_json.report_saved_raw_data
+            ):
+                metrics.append(parent_error_metric)
+        metrics = self.get_deduplicated_metrics(metrics=metrics)
+        return metrics
+
     @staticmethod
     def set_order_sex_for_sample(sample: Sample, metric_conditions: dict) -> None:
         metric_conditions["predicted_sex_sex_check"]["threshold"] = sample.sex

cg/models/deliverables/metric_deliverables.py

@@ -166,3 +166,4 @@ class MultiqcDataJson(BaseModel):
 
     report_general_stats_data: list[dict] | None
     report_data_sources: dict | None
+    report_saved_raw_data: dict[str, dict] | None

tests/fixtures/analysis/raredisease/raredisease_case_enough_reads_metrics_deliverables.yaml

@@ -286,7 +286,7 @@ metrics:
   value: 24.546493
 - condition:
     norm: gt
-    threshold: 26.0
+    threshold: 25.0
   header: null
   id: ADM1
   input: multiqc_data.json