feat: add qc step (#675)

* feat: add qc step * fix: remove .df * fix: fix in name * fix: fix v3 * Update src/gentropy/sumstat_qc_step.py Co-authored-by: Daniel Suveges <[email protected]> * Update src/gentropy/sumstat_qc_step.py Co-authored-by: Daniel Suveges <[email protected]> * fix: optimisation of code --------- Co-authored-by: Daniel Suveges <[email protected]>
opentargets · Jul 9, 2024 · 4a5f69f · 4a5f69f
1 parent 5f97232
commit 4a5f69f
Show file tree

Hide file tree

Showing 2 changed files with 47 additions and 0 deletions.
diff --git a/src/gentropy/config.py b/src/gentropy/config.py
@@ -435,6 +435,16 @@ class FinemapperConfig(StepConfig):
     _target_: str = "gentropy.susie_finemapper.SusieFineMapperStep"
 
 
+@dataclass
+class GWASQCStep(StepConfig):
+    """GWAS QC step configuration."""
+
+    gwas_path: str = MISSING
+    output_path: str = MISSING
+    studyid: str = MISSING
+    _target_: str = "gentropy.sumstat_qc_step.SummaryStatisticsQCStep"
+
+
 @dataclass
 class Config:
     """Application configuration."""

diff --git a/src/gentropy/sumstat_qc_step.py b/src/gentropy/sumstat_qc_step.py
@@ -0,0 +1,37 @@
+"""Step to calculate quality control metrics on the provided GWAS study."""
+
+from __future__ import annotations
+
+from gentropy.common.session import Session
+from gentropy.dataset.summary_statistics import SummaryStatistics
+from gentropy.method.sumstat_quality_controls import SummaryStatisticsQC
+
+
+class SummaryStatisticsQCStep:
+    """Step to run GWAS QC."""
+
+    def __init__(
+        self,
+        session: Session,
+        gwas_path: str,
+        output_path: str,
+        studyid: str,
+    ) -> None:
+        """Calculating quality control metrics on the provided GWAS study.
+
+        Args:
+            session (Session): Spark session
+            gwas_path (str): Path to the GWAS summary statistics.
+            output_path (str): Output path for the QC results.
+            studyid (str): Study ID for the QC.
+
+        """
+        gwas = SummaryStatistics.from_parquet(session, path=gwas_path)
+
+        (
+            SummaryStatisticsQC.get_quality_control_metrics(
+                gwas=gwas, limit=100_000_000, min_count=100, n_total=100000
+            )
+            .write.mode(session.write_mode)
+            .parquet(output_path + "/qc_results_" + studyid)
+        )