From 3d8330a2f60d3ab6e9c613128124aa4b461c0f35 Mon Sep 17 00:00:00 2001
From: Cunliang Geng <c.geng@esciencecenter.nl>
Date: Mon, 20 Nov 2023 11:46:01 +0100
Subject: [PATCH] remove function `filter_mibig_only_gcf`

This function is needed anymore. GCF loader already has the functionality to filter mibig only GCF during loading.
---
 src/nplinker/genomics/__init__.py | 2 --
 src/nplinker/genomics/genomics.py | 9 ---------
 src/nplinker/loader.py            | 3 +--
 tests/genomics/test_genomics.py   | 8 --------
 4 files changed, 1 insertion(+), 21 deletions(-)

diff --git a/src/nplinker/genomics/__init__.py b/src/nplinker/genomics/__init__.py
index 7710f337..cffff7c5 100644
--- a/src/nplinker/genomics/__init__.py
+++ b/src/nplinker/genomics/__init__.py
@@ -2,7 +2,6 @@
 from .abc import BGCLoaderBase
 from .bgc import BGC
 from .gcf import GCF
-from .genomics import filter_mibig_only_gcf
 from .genomics import generate_mappings_genome_id_bgc_id
 from .genomics import get_bgcs_from_gcfs
 from .genomics import get_strains_from_bgcs
@@ -16,7 +15,6 @@
     "BGCLoaderBase",
     "BGC",
     "GCF",
-    "filter_mibig_only_gcf",
     "generate_mappings_genome_id_bgc_id",
     "get_bgcs_from_gcfs",
     "get_strains_from_bgcs",
diff --git a/src/nplinker/genomics/genomics.py b/src/nplinker/genomics/genomics.py
index 53676be1..babadfe7 100644
--- a/src/nplinker/genomics/genomics.py
+++ b/src/nplinker/genomics/genomics.py
@@ -116,15 +116,6 @@ def map_bgc_to_gcf(bgcs: list[BGC], gcfs: list[GCF]):
             gcf.add_bgc(bgc)
 
 
-def filter_mibig_only_gcf(gcfs: list[GCF]) -> list[GCF]:
-    """Filter out GCFs that contain only MIBiG BGC objects.
-
-    This method returns a new list of GCFs that have at least one non-MIBiG
-    BGC object as its child.
-    """
-    return [gcf for gcf in gcfs if gcf.has_mibig_only() is False]
-
-
 def get_bgcs_from_gcfs(gcfs: list[GCF]) -> list[BGC]:
     """Get all BGC objects from given GCF objects."""
     s = set()
diff --git a/src/nplinker/loader.py b/src/nplinker/loader.py
index b6391c00..5155b723 100644
--- a/src/nplinker/loader.py
+++ b/src/nplinker/loader.py
@@ -8,7 +8,6 @@
 from nplinker.genomics import generate_mappings_genome_id_bgc_id
 from nplinker.genomics.antismash import AntismashBGCLoader
 from nplinker.genomics.bigscape import BigscapeGCFLoader
-from nplinker.genomics.genomics import filter_mibig_only_gcf
 from nplinker.genomics.genomics import get_bgcs_from_gcfs
 from nplinker.genomics.genomics import get_strains_from_bgcs
 from nplinker.genomics.genomics import map_bgc_to_gcf
@@ -453,7 +452,7 @@ def _load_genomics(self):
         map_bgc_to_gcf(raw_bgcs, raw_gcfs)
 
         # Step 5: get clean GCF objects, BGC objects and Strain objects
-        self.gcfs = filter_mibig_only_gcf(raw_gcfs)
+        self.gcfs = raw_gcfs
         self.bgcs = get_bgcs_from_gcfs(self.gcfs)
         self.strains = get_strains_from_bgcs(self.bgcs)
 
diff --git a/tests/genomics/test_genomics.py b/tests/genomics/test_genomics.py
index f691d561..394f5a36 100644
--- a/tests/genomics/test_genomics.py
+++ b/tests/genomics/test_genomics.py
@@ -3,7 +3,6 @@
 import pytest
 from nplinker.genomics import BGC
 from nplinker.genomics import GCF
-from nplinker.genomics import filter_mibig_only_gcf
 from nplinker.genomics import generate_mappings_genome_id_bgc_id
 from nplinker.genomics import get_bgcs_from_gcfs
 from nplinker.genomics import get_strains_from_bgcs
@@ -149,13 +148,6 @@ def test_map_bgc_to_gcf_error(bgc_list, gcf_list_error):
     assert "BGC id 'BGC_04' from GCF object '1' not found" in e.value.args[0]
 
 
-def test_filter_mibig_only_gcf(bgc_list, gcf_list):
-    map_bgc_to_gcf(bgc_list, gcf_list)
-    gcfs = filter_mibig_only_gcf(gcf_list)
-    assert len(gcfs) == 1
-    assert gcfs[0].gcf_id == "2"
-
-
 def test_get_bgcs_from_gcfs(bgc_list, gcf_list):
     map_bgc_to_gcf(bgc_list, gcf_list)
     bgcs = get_bgcs_from_gcfs(gcf_list)