Merge pull request #36 from genotoul-bioinfo/dev

Fix missing score in input bins

binette/bin_manager.py

@@ -342,26 +342,19 @@ def get_bins_from_contig2bin_table(contig2bin_table: Path, set_name: str) -> Lis
     return bins
 
 
-def from_bin_sets_to_bin_graph(
-    bin_name_to_bin_set: Mapping[str, Iterable[Bin]]
-) -> nx.Graph:
+def from_bins_to_bin_graph(bins) -> nx.Graph:
     """
-    Creates a bin graph from a dictionary of bin sets.
+    Creates a bin graph made of overlapping gram a set of bins.
 
-    :param bin_name_to_bin_set: A dictionary mapping bin names to their respective bin sets.
+    :param bins: a set of bins
 
     :return: A networkx Graph representing the bin graph of overlapping bins.
     """
     G = nx.Graph()
 
-    for set1_name, set2_name in itertools.combinations(bin_name_to_bin_set, 2):
-        set1 = bin_name_to_bin_set[set1_name]
-        set2 = bin_name_to_bin_set[set2_name]
-
-        for bin1, bin2 in itertools.product(set1, set2):
-
-            if bin1.overlaps_with(bin2):
-                G.add_edge(bin1, bin2)
+    for bin1, bin2 in itertools.combinations(bins, 2):
+        if bin1.overlaps_with(bin2):
+            G.add_edge(bin1, bin2)
     return G
 
 
@@ -618,18 +611,16 @@ def rename_bin_contigs(bins: Iterable[Bin], contig_to_index: dict):
         b.hash = hash(str(sorted(b.contigs)))
 
 
-def create_intermediate_bins(
-    bin_set_name_to_bins: Mapping[str, Iterable[Bin]]
-) -> Set[Bin]:
+def create_intermediate_bins(original_bins: Set[Bin]) -> Set[Bin]:
     """
     Creates intermediate bins from a dictionary of bin sets.
 
-    :param bin_set_name_to_bins: A dictionary mapping bin set names to corresponding bins.
+    :param original_bins: Set of input bins.
 
     :return: A set of intermediate bins created from intersections, differences, and unions.
     """
     logging.info("Making bin graph...")
-    connected_bins_graph = from_bin_sets_to_bin_graph(bin_set_name_to_bins)
+    connected_bins_graph = from_bins_to_bin_graph(original_bins)
 
     logging.info("Creating intersection bins...")
     intersection_bins = get_intersection_bins(connected_bins_graph)

binette/bin_quality.py

@@ -242,7 +242,7 @@ def assess_bins_quality_by_chunk(
 
     for i, chunk_bins_iter in enumerate(chunks(bins, chunk_size)):
         chunk_bins = set(chunk_bins_iter)
-        logging.debug(f"chunk {i}: assessing quality of {len(chunk_bins)}")
+        logging.debug(f"chunk {i}: assessing quality of {len(chunk_bins)} bins")
         assess_bins_quality(
             bins=chunk_bins,
             contig_to_kegg_counter=contig_to_kegg_counter,

binette/io_manager.py

@@ -1,3 +1,4 @@
+from collections import defaultdict
 import logging
 import pyfastx
 from typing import Iterable, List, Dict, Tuple, Set
@@ -234,23 +235,24 @@ def check_resume_file(faa_file: Path, diamond_result_file: Path) -> None:
         raise FileNotFoundError(error_msg)
 
 
-def write_original_bin_metrics(
-    bin_set_name_to_bins: Dict[str, Set[Bin]], original_bin_report_dir: Path
-):
+def write_original_bin_metrics(original_bins: Set[Bin], original_bin_report_dir: Path):
     """
     Write metrics of original input bins to a specified directory.
 
-    This function takes a dictionary mapping bin set names to sets of bins and writes
-    the metrics for each bin set to a TSV file in the specified directory. Each bin set
-    will have its own TSV file named according to its set name.
+    This function writes the metrics for each bin set to a TSV file in the specified directory.
+    Each bin set will have its own TSV file named according to its set name.
 
-    :param bin_set_name_to_bins: A dictionary where the keys are bin set names (str) and
-                                 the values are sets of Bin objects representing bins.
+    :param original_bins: A set containing input bins
     :param original_bin_report_dir: The directory path (Path) where the bin metrics will be saved.
     """
 
     original_bin_report_dir.mkdir(parents=True, exist_ok=True)
 
+    bin_set_name_to_bins = defaultdict(set)
+    for bin_obj in original_bins:
+        for origin in bin_obj.origin:
+            bin_set_name_to_bins[origin].add(bin_obj)
+
     for i, (set_name, bins) in enumerate(sorted(bin_set_name_to_bins.items())):
         bins_metric_file = (
             original_bin_report_dir

binette/main.py

@@ -226,7 +226,6 @@ def parse_input_files(
     :fasta_extensions: Possible fasta extensions to look for in the bin directory.
 
     :return: A tuple containing:
-        - Dictionary mapping bin set names to lists of bins.
         - List of original bins.
         - Dictionary mapping bins to lists of contigs.
         - Dictionary mapping contig names to their lengths.
@@ -271,7 +270,7 @@ def parse_input_files(
         seq.name: len(seq) for seq in contigs_object if seq.name in contigs_in_bins
     }
 
-    return bin_set_name_to_bins, original_bins, contigs_in_bins, contig_to_length
+    return original_bins, contigs_in_bins, contig_to_length
 
 
 def manage_protein_alignement(
@@ -500,13 +499,11 @@ def main():
         io.check_resume_file(faa_file, diamond_result_file)
         use_existing_protein_file = True
 
-    bin_set_name_to_bins, original_bins, contigs_in_bins, contig_to_length = (
-        parse_input_files(
-            args.bin_dirs,
-            args.contig2bin_tables,
-            args.contigs,
-            fasta_extensions=set(args.fasta_extensions),
-        )
+    original_bins, contigs_in_bins, contig_to_length = parse_input_files(
+        args.bin_dirs,
+        args.contig2bin_tables,
+        args.contigs,
+        fasta_extensions=set(args.fasta_extensions),
     )
 
     contig_to_kegg_counter, contig_to_genes = manage_protein_alignement(
@@ -552,10 +549,10 @@ def main():
     logging.info(
         f"Writting original input bin metrics to directory: {original_bin_report_dir}"
     )
-    io.write_original_bin_metrics(bin_set_name_to_bins, original_bin_report_dir)
+    io.write_original_bin_metrics(original_bins, original_bin_report_dir)
 
     logging.info("Create intermediate bins:")
-    new_bins = bin_manager.create_intermediate_bins(bin_set_name_to_bins)
+    new_bins = bin_manager.create_intermediate_bins(original_bins)
 
     logging.info("Assess quality for supplementary intermediate bins.")
     new_bins = bin_quality.add_bin_metrics(

tests/bin_manager_test.py

@@ -256,11 +256,11 @@ def test_select_best_bins_with_equality():
 
 # The function should create intersection bins when there are overlapping contigs between bins.
 def test_intersection_bins_created():
-    set1 = [
+    set1 = {
         bin_manager.Bin(contigs={"1", "2"}, origin="A", name="bin1"),
         bin_manager.Bin(contigs={"3", "4"}, origin="A", name="bin2"),
         bin_manager.Bin(contigs={"5"}, origin="A", name="bin2"),
-    ]
+    }
     # need to defined completeness and conta
     # because when too low the bin is not used in all operation
     for b in set1:
@@ -270,14 +270,12 @@ def test_intersection_bins_created():
     binA = bin_manager.Bin(contigs={"1", "3"}, origin="B", name="binA")
     binA.contamination = 0
     binA.completeness = 100
-    set2 = [
+    set2 = {
         binA,
-    ]
-    bin_set_name_to_bins = {"set1": set1, "set2": set2}
+    }
+    input_bins = set1 | set2
 
-    intermediate_bins_result = bin_manager.create_intermediate_bins(
-        bin_set_name_to_bins
-    )
+    intermediate_bins_result = bin_manager.create_intermediate_bins(input_bins)
 
     expected_intermediate_bins = {
         bin_manager.Bin(contigs={"1", "2", "3"}, origin="bin1 | binA ", name="NA"),
@@ -337,19 +335,19 @@ def test_dereplicate_bin_sets():
     assert derep_bins_result == {b1, b2, b3}
 
 
-def test_from_bin_sets_to_bin_graph():
+def test_from_bins_to_bin_graph():
 
     bin1 = bin_manager.Bin(contigs={"1", "2"}, origin="A", name="bin1")
     bin2 = bin_manager.Bin(contigs={"3", "4"}, origin="A", name="bin2")
     bin3 = bin_manager.Bin(contigs={"5"}, origin="A", name="bin3")
 
-    set1 = [bin1, bin2, bin3]
+    set1 = {bin1, bin2, bin3}
 
     binA = bin_manager.Bin(contigs={"1", "3"}, origin="B", name="binA")
 
-    set2 = [binA]
+    set2 = {binA}
 
-    result_graph = bin_manager.from_bin_sets_to_bin_graph({"B": set2, "A": set1})
+    result_graph = bin_manager.from_bins_to_bin_graph(set1 | set2)
 
     assert result_graph.number_of_edges() == 2
     # bin3 is not connected to any bin so it is not in the graph

tests/io_manager_test.py

@@ -350,17 +350,16 @@ def test_write_original_bin_metrics(mock_write_bin_info, bin1, bin2, tmp_path):
 
     temp_directory = tmp_path / "test_output"
 
-    mock_bins = {"set1": {bin1}, "set2": {bin2}}
     # Call the function with mock data
-    io_manager.write_original_bin_metrics(mock_bins, temp_directory)
+    io_manager.write_original_bin_metrics({bin1, bin2}, temp_directory)
 
     # Check if the output directory was created
     assert temp_directory.exists(), "Output directory should be created."
 
     # Check that the correct files are created
     expected_files = [
-        temp_directory / "input_bins_1.set1.tsv",
-        temp_directory / "input_bins_2.set2.tsv",
+        temp_directory / "input_bins_1.origin1.tsv",
+        temp_directory / "input_bins_2.origin2.tsv",
     ]
 
     assert (
@@ -373,5 +372,5 @@ def test_write_original_bin_metrics(mock_write_bin_info, bin1, bin2, tmp_path):
     ), "write_bin_info should be called once for each bin set."
 
     # Verify the specific calls to `write_bin_info`
-    mock_write_bin_info.assert_any_call(mock_bins["set1"], expected_files[0])
-    mock_write_bin_info.assert_any_call(mock_bins["set2"], expected_files[1])
+    mock_write_bin_info.assert_any_call({bin1}, expected_files[0])
+    mock_write_bin_info.assert_any_call({bin2}, expected_files[1])

tests/main_binette_test.py

@@ -207,17 +207,15 @@ def test_parse_input_files_with_contig2bin_tables(tmp_path):
     fasta_file.write_text(fasta_file_content)
 
     # Call the function and capture the return values
-    bin_set_name_to_bins, original_bins, contigs_in_bins, contig_to_length = (
-        parse_input_files(None, [bin_set1, bin_set2], fasta_file)
+    original_bins, contigs_in_bins, contig_to_length = parse_input_files(
+        None, [bin_set1, bin_set2], fasta_file
     )
 
     # # Perform assertions on the returned values
-    assert isinstance(bin_set_name_to_bins, dict)
     assert isinstance(original_bins, set)
     assert isinstance(contigs_in_bins, set)
     assert isinstance(contig_to_length, dict)
 
-    assert set(bin_set_name_to_bins) == {"bin_set1", "bin_set2"}
     assert len(original_bins) == 4
     assert contigs_in_bins == {"contig1", "contig2", "contig3", "contig4"}
     assert len(contig_to_length) == 4
@@ -248,17 +246,15 @@ def test_parse_input_files_bin_dirs(create_temp_bin_directories, tmp_path):
     fasta_file.write_text(fasta_file_content)
 
     # Call the function and capture the return values
-    bin_set_name_to_bins, original_bins, contigs_in_bins, contig_to_length = (
-        parse_input_files(bin_dirs, contig2bin_tables, fasta_file)
+    original_bins, contigs_in_bins, contig_to_length = parse_input_files(
+        bin_dirs, contig2bin_tables, fasta_file
     )
 
     # # Perform assertions on the returned values
-    assert isinstance(bin_set_name_to_bins, dict)
     assert isinstance(original_bins, set)
     assert isinstance(contigs_in_bins, set)
     assert isinstance(contig_to_length, dict)
 
-    assert set(bin_set_name_to_bins) == {"set1", "set2"}
     assert len(original_bins) == 3
     assert contigs_in_bins == {
         "contig1",
@@ -471,7 +467,7 @@ def test_main(monkeypatch, test_environment):
     ) as mock_select_bins_and_write_them:
 
         # Set return values for mocked functions if needed
-        mock_parse_input_files.return_value = (None, None, None, None)
+        mock_parse_input_files.return_value = (None, None, None)
         mock_manage_protein_alignement.return_value = (
             {"contig1": 1},
             {"contig1": ["gene1"]},