Merge pull request #17 from nkongenelly/DATAOPS-832_olink_cluster_info

Implementing total_clusters_raw similar to as total_clusters_pf

Author: nkongenelly

projman_filler/interop_run_stats_parser.py

@@ -114,24 +114,18 @@ def _get_non_index_reads(self) -> list:
     def _get_conversion_results(self) -> list:
         ar = iop.summary(self._run_metrics, 'Lane')
         df = pd.DataFrame(ar)
-
         # Get statistics per-lane
         n_lanes = self._run_summary.lane_count()
         lanes = []
+
         for l in range(1, n_lanes+1):
             rows = df.loc[df['Lane'] == l][['ReadNumber','Reads', 'Reads Pf', 'IsIndex']]
             rows = rows.reset_index()
-            # Each 'Reads Pf' value represents the entire lane,
-            # while 'Reads' (total clusters raw) must be summed over the non-index reads
-            # See: https://github.com/Illumina/interop/issues/271
-            total_clusters_raw = 0
-            total_clusters_pf = 0
-            for index, row in rows.iterrows():
-                if index in self._non_index_reads:
-                    # These are the same for the lane across all reads
-                    total_clusters_pf = row['Reads Pf']
-                    # These must be summed
-                    total_clusters_raw += row['Reads']
+            
+            # These are the same for the lane across all reads
+            total_clusters_pf = rows.iloc[0].get('Reads Pf', 0)
+            total_clusters_raw = rows.iloc[0].get('Reads', 0)
+            
             lanes.append(Lane(l, total_clusters_raw, total_clusters_pf))
         return lanes
 

tests/test_interop_run_stats_parser.py

@@ -2,29 +2,68 @@
 from unittest.mock import MagicMock
 from projman_filler.interop_run_stats_parser import InteropRunStatsParser
 from interop import py_interop_run, py_interop_run_metrics, py_interop_summary
+import pandas as pd
 import interop
+from projman_filler.lane import Lane
+
 
 class TestRunStatsParsers(unittest.TestCase):
     def test_interop_standardize_read_numbers(self):
         runfolder = "tests/resources/200624_A00834_0183_BHMTFYTINY"
 
-        non_index_reads = [0, 2, 3]
+        non_index_reads = [0]
         iop = InteropRunStatsParser(runfolder, non_index_reads)
         remapped = iop._standardize_read_numbers()
-        expected = {1: 0, 2: 2, 3: 3}
+        expected = {1: 0}
         self.assertEqual(remapped, expected)
 
         reads = iop.get_reads()
-        expected = [1, 2, 3]
+        expected = [1]
         self.assertEqual(reads, expected)
 
     def test_lanes_total_clusters(self):
-        non_index_reads = [0, 2, 3]
+        non_index_reads = [0]
         runfolder = "tests/resources/200624_A00834_0183_BHMTFYTINY"
         iop = InteropRunStatsParser(runfolder, non_index_reads)
         for lane in iop._conversion_results:
             assert lane._total_clusters_pf is not None and lane._total_clusters_pf != 0
             assert lane._total_clusters_raw is not None and lane._total_clusters_raw != 0
 
+    def test_clusters_same_across_lanes(self):
+        """
+        Verify that 'Reads' and 'Reads Pf' are consistently the same across all reads within the same lane
+        """
+        non_index_reads = [0]
+        runfolder = "tests/resources/200624_A00834_0183_BHMTFYTINY"
+        iop = InteropRunStatsParser(runfolder, non_index_reads)
+        
+        interop_lane_summary = interop.summary(iop._run_metrics, 'Lane')
+        df = pd.DataFrame(interop_lane_summary)
+
+        for lane_index, (lane, rows) in enumerate(df.groupby('Lane')):
+            rows = rows.reset_index()
+            # Assert all 'Reads' and 'Reads Pf' values are consistent within the lane
+            assert rows['Reads'].nunique() == 1, (
+                    f"Inconsistent 'Reads' in lane {lane}"
+                    )
+            assert rows['Reads Pf'].nunique() == 1, (
+                    f"Inconsistent 'Reads Pf' in lane {lane}"
+                    )
+
+            # These are the same for the lane across all reads
+            total_clusters_pf = rows.at[0, 'Reads Pf']
+            total_clusters_raw = rows.at[0, 'Reads']
+            
+            
+            # Compare with InteropRunStatsParser results
+            lane_results = iop._conversion_results[lane_index]
+            assert lane_results._total_clusters_pf == total_clusters_pf, (
+                    f"Mismatch in total_clusters_pf for lane {lane}"
+                    )
+            assert lane_results._total_clusters_raw == total_clusters_raw, (
+                    f"Mismatch in total_clusters_raw for lane {lane}"
+                    )
+                
 if __name__ == '__main__':
     unittest.main()
+