From 6a8155ec0b227da9907fb709c909afdb35571375 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ann-Kathrin=20Br=C3=BCggemann?=
 <90249112+AKBrueggemann@users.noreply.github.com>
Date: Tue, 4 Jan 2022 08:43:04 +0100
Subject: [PATCH] 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 <46334240+thomasbtf@users.noreply.github.com>
---
 workflow/scripts/plot-lineages-over-time.py | 15 +++++++++++----
 workflow/scripts/plot-variants-over-time.py | 14 ++++++++++----
 2 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/workflow/scripts/plot-lineages-over-time.py b/workflow/scripts/plot-lineages-over-time.py
index bea202b12..17fd019f3 100644
--- 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
index 4be2b776a..7520402cc 100644
--- 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)