From 42dd8086e5ff4a54e7b74308789287fc1c112771 Mon Sep 17 00:00:00 2001
From: Ivan Blagoev Topolsky <ivan.topolsky@sib.swiss>
Date: Wed, 3 Jul 2024 18:11:46 +0200
Subject: [PATCH] HACK: remove filtering

- replace with something generic in the future
---
 lollipop/preprocessors.py | 73 ++-------------------------------------
 1 file changed, 3 insertions(+), 70 deletions(-)

diff --git a/lollipop/preprocessors.py b/lollipop/preprocessors.py
index 07d83b6..017abdc 100644
--- a/lollipop/preprocessors.py
+++ b/lollipop/preprocessors.py
@@ -116,75 +116,8 @@ def general_preprocess(
 
         return self
 
-    def filter_mutations(self):
-        """very temporary function, to filter out hardcoded problematic mutations"""
-
-        # HACK hand hardcoded
-        # TODO detect problematic mutations
-        self.df_tally = self.df_tally[
-            ~self.df_tally["mutations"].isin(
-                ["28461G", "11201G", "26801C"] + ["-28461G", "-11201G", "-26801C"]
-            )
-        ]
-
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos >= 22428)
-                & (self.df_tally.pos <= 22785)
-            )
-        ]  # amplicon75
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos >= 22677)
-                & (self.df_tally.pos <= 23028)
-            )
-        ]  # amplicon76
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos >= 22974)
-                & (self.df_tally.pos <= 23327)
-            )
-        ]  # amplicon77
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos >= 26277)
-                & (self.df_tally.pos <= 26635)
-            )
-        ]  # amplicon88
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos >= 26895)
-                & (self.df_tally.pos <= 27256)
-            )
-        ]  # amplicon90
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos == 26709)
-            )
-        ]  # other
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos == 27807)
-            )
-        ]  # other
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos == 2832)
-            )
-        ]  # other
-        self.df_tally = self.df_tally[
-            ~(
-                (pd.to_datetime(self.df_tally.date) > np.datetime64("2021-11-20"))
-                & (self.df_tally.pos == 10449)
-            )
-        ]  # other
+    def filter_mutations(self, filters=None):
+        """filter out hardcoded problematic mutations"""
 
+        # HACK completely disable filters
         return self