fix: relative aggregation sample plot (#412)

* Changed plots occurence threshold to relative * fmt * Name changed: number of occurences in plot * fmt * Changed threshold values for variant/lineage plots * Changed way of thresholding lineages/variants * fmt * Changed name for other occ. in plot Co-authored-by: Thomas Battenfeld <[email protected]>
IKIM-Essen · Jan 4, 2022 · 6a8155e · 6a8155e
1 parent 0e8f993
commit 6a8155e
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 8 deletions.
diff --git a/workflow/scripts/plot-lineages-over-time.py b/workflow/scripts/plot-lineages-over-time.py
@@ -34,10 +34,17 @@ def plot_lineages_over_time(sm_input, sm_output, dates, sm_output_table):
         pangolin_calls["lineage_count"] = pd.Series()
 
     # mask low occurrences
-    threshold = len(pangolin_calls) / 10
-    pangolin_calls.loc[pangolin_calls["lineage_count"] < threshold, "lineage"] = (
-        "other (<" + str(threshold) + " occ.)"
-    )
+    print(pangolin_calls["lineage"].value_counts())
+    df = pd.DataFrame(pangolin_calls["lineage"].value_counts())
+    df.sort_values(by=["lineage"])
+    if len(df.index) > 10:
+        pangolin_calls.loc[
+            ~df.head(10).isin(pangolin_calls["lineage"]), "lineage"
+        ] = "other occ."
+    else:
+        pangolin_calls.loc[
+            pangolin_calls["lineage_count"] < 0, "lineage"
+        ] = "other occ."
 
     pangolin_calls.rename(columns={"lineage": "Lineage", "date": "Date"}, inplace=True)
 

diff --git a/workflow/scripts/plot-variants-over-time.py b/workflow/scripts/plot-variants-over-time.py
@@ -84,10 +84,16 @@ def plot_variants_over_time(sm_output, sm_output_table):
         ].transform(lambda s: s.count())
 
         # mask low occurrences
-        threshold = len(calls) / 10
-        calls.loc[calls["total occurrence"] < threshold, "alteration"] = (
-            "other (<" + str(threshold) + " occ.)"
-        )
+        print(calls["alteration"].value_counts())
+        df = pd.DataFrame(calls["alteration"].value_counts())
+        df.sort_values(by=["alteration"])
+        if len(df.index) > 10:
+            # print(calls.loc[calls["alteration"].isin(df.head(10).index)])
+            calls.loc[
+                ~calls["alteration"].isin(df.head(10).index), "alteration"
+            ] = "other occ."
+        else:
+            calls.loc[calls["total occurrence"] < 0, "alteration"] = "other occ."
 
     calls.rename(columns={"alteration": "Alteration", "date": "Date"}, inplace=True)