From e0d536687eb26d5071f8618fba492718cd70a73c Mon Sep 17 00:00:00 2001
From: Arjan Draisma <arjan.draisma@wur.nl>
Date: Tue, 22 Aug 2023 10:52:46 +0200
Subject: [PATCH] removed sortedlist typing fixes

---
 bigscape.py        | 38 ++++++++++++++++----------------------
 src/genbank/cds.py | 13 ++++---------
 src/hmm/hsp.py     |  5 +----
 3 files changed, 21 insertions(+), 35 deletions(-)

diff --git a/bigscape.py b/bigscape.py
index 334b0eba..ea576759 100644
--- a/bigscape.py
+++ b/bigscape.py
@@ -51,7 +51,7 @@ def load_data(run: RunParameters):
         profiler = Profiler(run.output.profile_path)
         profiler.start()
 
-    if run.output.db_path.exists():
+    if False:
         logging.info("Loading existing run from disk...")
 
         DB.load_from_disk(run.output.db_path)
@@ -60,9 +60,6 @@ def load_data(run: RunParameters):
 
         for gbk in gbks:
             HSP.load_all(gbk.genes)
-            for cds in gbk.genes:
-                # TODO: remove once we get rid of sortedlists
-                cds.lock()
 
         HMMer.init(run.input.pfam_path)
 
@@ -110,7 +107,7 @@ def callback(tasks_done):
 
         if platform.system() == "Darwin":
             logging.warning("Running on mac-OS: hmmsearch_simple single threaded")
-            all_hsps = list(HMMer.hmmsearch_simple(all_cds, 1))
+            HMMer.hmmsearch_simple(all_cds, 1)
         else:
             logging.debug(
                 "Running on %s: hmmsearch_multiprocess with %d cores",
@@ -118,30 +115,25 @@ def callback(tasks_done):
                 run.cores,
             )
 
-            # if legacy is true, set cutoff to 1.1 for the domain filtering so we can use
-            # legacy filtering later
-
-            if run.legacy:
-                domain_overlap_cutoff = 1.1
-            else:
-                domain_overlap_cutoff = run.hmmer.domain_overlap_cutoff
+            # TODO: the overlap filtering in this function does not seem to work
             HMMer.hmmsearch_multiprocess(
                 all_cds,
-                domain_overlap_cutoff=domain_overlap_cutoff,
+                domain_overlap_cutoff=run.hmmer.domain_overlap_cutoff,
                 cores=run.cores,
             )
 
         # TODO: move, or remove after the add_hsp_overlap function is fixed (if it is broken
         # in the first place)
-        for cds in all_cds:
-            if run.legacy:
-                cds.hsps = legacy_filter_overlap(cds.hsps, 0.1)
-
-            # TODO: remove when sortedlists are removed
-            # this converts the sortedlist used internally to regular lists.
-            # for some reason, doing this beforehand really messes things up for reasons I don't
-            # understand.
-            cds.lock()
+        # this sorts all CDS and then filters them using the old filtering system, which
+        # is less efficient than the flitering using the CDS.add_hsp_overlap_filter
+        # method. however, that method seems to be broken somehow
+        all_hsps = []
+        for gbk in gbks:
+            for cds in gbk.genes:
+                cds.hsps = sorted(cds.hsps)
+                all_hsps.extend(
+                    [hsp.domain for hsp in legacy_filter_overlap(cds.hsps, 0.1)]
+                )
 
         all_hsps = []
         for cds in all_cds:
@@ -171,6 +163,8 @@ def callback(tasks_done):
         all_alignments = list()
         for cds in all_cds:
             for hsp in cds.hsps:
+                if hsp.alignment is None:
+                    continue
                 all_alignments.append(hsp.alignment)
 
         logging.info("%d alignments", len(all_alignments))
diff --git a/src/genbank/cds.py b/src/genbank/cds.py
index a99b6ad0..c1872869 100644
--- a/src/genbank/cds.py
+++ b/src/genbank/cds.py
@@ -7,7 +7,6 @@
 
 # from dependencies
 from Bio.SeqFeature import SeqFeature
-from sortedcontainers import SortedList
 
 # from other modules
 from src.errors import InvalidGBKError
@@ -32,7 +31,7 @@ class CDS:
         gene_kind: str
         strand: Bool
         aa_seq: SeqRecord.seq
-        hsps: SortedList[HSP]
+        hsps: list[HSP]
     """
 
     def __init__(self, nt_start: int, nt_stop: int):
@@ -43,7 +42,7 @@ def __init__(self, nt_start: int, nt_stop: int):
         self.gene_kind: Optional[str] = None
         self.strand: Optional[int] = None
         self.aa_seq: str = ""
-        self.hsps: SortedList[HSP] | list[HSP] = SortedList()
+        self.hsps: list[HSP] = []
         self.__locked = False
 
         # db specific fields
@@ -65,13 +64,9 @@ def add_hsp_overlap_filter(self, new_hsp: HSP, domain_overlap_cutoff=0.1) -> Non
             0.1
 
         """
-        # TODO: remove once sorted lists are gone entirely
-        if self.__locked or isinstance(self.hsps, list):
-            raise AttributeError("Cannot add HSPs to a locked CDS")
-
         # if no hsps added yet, just add and continue
         if len(self.hsps) == 0:
-            self.hsps.add(new_hsp)
+            self.hsps.append(new_hsp)
             return
 
         delete_list = []
@@ -117,7 +112,7 @@ def add_hsp_overlap_filter(self, new_hsp: HSP, domain_overlap_cutoff=0.1) -> Non
 
         # if we got through all of that without the function, we never replaced an HSP
         # so add a new one here
-        self.hsps.add(new_hsp)
+        self.hsps.append(new_hsp)
 
     def save(self, commit=True):
         """Saves this CDS to the database and optionally executes a commit
diff --git a/src/hmm/hsp.py b/src/hmm/hsp.py
index ff862d1f..670904cf 100644
--- a/src/hmm/hsp.py
+++ b/src/hmm/hsp.py
@@ -163,10 +163,7 @@ def load_all(cds_list: list[CDS]) -> None:
             )
             new_hsp.alignment = HSPAlignment(new_hsp, result.alignment)
 
-            if isinstance(cds.hsps, list):
-                raise ValueError("HSP list of CDS is a list. Did the CDS get locked?")
-
-            cds.hsps.add(new_hsp)
+            cds.hsps.append(new_hsp)
 
     @staticmethod
     def has_overlap(hsp_a: HSP, hsp_b: HSP) -> bool: