added assembly attributes to API; lots of associated changes

maintenance/main.py

@@ -302,20 +302,50 @@ def write_genome_data(self, ensembl_id):
             )
 
         genome_data_path = path / "genome_data.py"
-        # Check if file exists and was manually created
+        # Check if file exists and has non-Ensembl assembly source
         if genome_data_path.exists():
-            with open(genome_data_path) as f:
-                first_line = f.readline().strip()
-                if first_line.startswith("# File created manually"):
+            try:
+                # Get existing data
+                namespace = {}
+                with open(genome_data_path) as f:
+                    exec(f.read(), namespace)
+                existing_assembly_source = namespace["data"].get(
+                    "assembly_source", "ensembl"
+                )
+                if existing_assembly_source != "ensembl":
                     logger.info(
-                        f"Skipping {sps_id} ({ensembl_id}): manually created \
-                            genome data file"
+                        f"Skipping {sps_id} ({ensembl_id}): non-Ensembl assembly source"
                     )
                     return ("manual", None)
+            except Exception as e:
+                logger.warning(
+                    f"Error checking assembly source for {sps_id}: {e}. "
+                    "Proceeding with update."
+                )
 
         # Get new data from Ensembl
         data = self.ensembl_client.get_genome_data(ensembl_id)
 
+        # Preserve existing assembly source or default to "ensembl"
+        if genome_data_path.exists():
+            try:
+                namespace = {}
+                with open(genome_data_path) as f:
+                    exec(f.read(), namespace)
+                data["assembly_source"] = namespace["data"].get(
+                    "assembly_source", "ensembl"
+                )
+            except Exception:
+                data["assembly_source"] = "ensembl"
+        else:
+            data["assembly_source"] = "ensembl"
+
+        # Add Ensembl version number if assembly source is Ensembl
+        if data["assembly_source"] == "ensembl":
+            data["assembly_build_version"] = str(self.ensembl_client.get_release())
+        else:
+            data["assembly_build_version"] = None
+
         # Check if existing genome data exists and compare chromosome names
         if genome_data_path.exists():
             try:
@@ -455,7 +485,7 @@ def update_genome_data(species):
             logger.error(f"Unexpected error processing {eid}: {e}")
             skipped_species.append((species_id, eid, f"Unexpected error: {str(e)}"))
 
-    writer.write_ensembl_release()
+    # writer.write_ensembl_release()
 
     # Add summary report at the end
     if skipped_species:
@@ -484,7 +514,7 @@ def add_species(ensembl_id, force):
     """
     writer = DataWriter()
     writer.add_species(ensembl_id.lower(), force=force)
-    writer.write_ensembl_release()
+    # writer.write_ensembl_release()
 
 
 # TODO refactor this so that it's an option to add-species. By default

stdpopsim/catalog/AedAeg/genome_data.py

@@ -8,4 +8,6 @@
         "3": {"length": 409777670, "synonyms": []},
         "MT": {"length": 16790, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/AnaPla/genome_data.py

@@ -45,4 +45,6 @@
         "39": {"length": 2018729, "synonyms": []},
         "40": {"length": 1354177, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/AnoCar/genome_data.py

@@ -18,4 +18,6 @@
         "LGh": {"length": 248369, "synonyms": []},
         "MT": {"length": 17223, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "103",
 }

stdpopsim/catalog/AnoGam/genome_data.py

@@ -10,4 +10,6 @@
         "X": {"length": 24393108, "synonyms": []},
         "Mt": {"length": 15363, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/ApiMel/genome_data.py

@@ -21,4 +21,6 @@
         "CM009946.2": {"length": 7238532, "synonyms": []},
         "CM009947.2": {"length": 16343, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/AraTha/genome_data.py

@@ -11,4 +11,6 @@
         "Mt": {"length": 366924, "synonyms": []},
         "Pt": {"length": 154478, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/BosTau/genome_data.py

@@ -35,4 +35,6 @@
         "X": {"length": 139009144, "synonyms": ["chrX"]},
         "MT": {"length": 16338, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/BosTau/species.py

@@ -59,6 +59,12 @@
 # 24.35 / 2628394923 = 9.26e-9 per bp per generation.
 _genome_wide_recombination_rate = 9.26e-9
 
+# Mutation rate
+_mutation_rate = 1.2e-8
+_mutation_rate_data = {str(i): _mutation_rate for i in range(1, 30)}
+_mutation_rate_data["MT"] = _mutation_rate
+_mutation_rate_data["X"] = _mutation_rate
+
 _recombination_rate_data = collections.defaultdict(
     lambda: _genome_wide_recombination_rate
 )
@@ -67,39 +73,8 @@
 
 # Generic and chromosome-specific ploidy
 _species_ploidy = 2
-_ploidy = {
-    "1": _species_ploidy,
-    "2": _species_ploidy,
-    "3": _species_ploidy,
-    "4": _species_ploidy,
-    "5": _species_ploidy,
-    "6": _species_ploidy,
-    "7": _species_ploidy,
-    "8": _species_ploidy,
-    "9": _species_ploidy,
-    "10": _species_ploidy,
-    "11": _species_ploidy,
-    "12": _species_ploidy,
-    "13": _species_ploidy,
-    "14": _species_ploidy,
-    "15": _species_ploidy,
-    "16": _species_ploidy,
-    "17": _species_ploidy,
-    "18": _species_ploidy,
-    "19": _species_ploidy,
-    "20": _species_ploidy,
-    "21": _species_ploidy,
-    "22": _species_ploidy,
-    "23": _species_ploidy,
-    "24": _species_ploidy,
-    "25": _species_ploidy,
-    "26": _species_ploidy,
-    "27": _species_ploidy,
-    "28": _species_ploidy,
-    "29": _species_ploidy,
-    "X": _species_ploidy,
-    "MT": 1,
-}
+_ploidy = {str(i): _species_ploidy for i in range(1, 30)}
+_ploidy.update({"X": _species_ploidy, "MT": 1})
 
 _chromosomes = []
 for name, data in genome_data.data["chromosomes"].items():
@@ -115,8 +90,11 @@
         )
     )
 
-_genome = stdpopsim.Genome(
-    chromosomes=_chromosomes,
+_genome = stdpopsim.Genome.from_data(
+    genome_data=genome_data.data,
+    recombination_rate=_recombination_rate_data,
+    mutation_rate=_mutation_rate_data,
+    ploidy=_ploidy,
     citations=[
         _HoweEtAl,  # ASSEMBLY
         _RosenEtAl,  # ASSEMBLY

stdpopsim/catalog/CaeEle/genome_data.py

@@ -11,4 +11,6 @@
         "X": {"length": 17718942, "synonyms": []},
         "MtDNA": {"length": 13794, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/CaeEle/species.py

@@ -123,10 +123,11 @@
         )
     )
 
-_genome = stdpopsim.Genome(
-    chromosomes=_chromosomes,
-    assembly_name=genome_data.data["assembly_name"],
-    assembly_accession=genome_data.data["assembly_accession"],
+_genome = stdpopsim.Genome.from_data(
+    genome_data=genome_data.data,
+    recombination_rate=_recombination_rate_data,
+    mutation_rate=_mutation_rate_data,
+    ploidy=_ploidy,
     citations=[
         _genome1998,
         _KonradEtAl2019.because(stdpopsim.CiteReason.MUT_RATE),

stdpopsim/catalog/CanFam/genome_data.py

@@ -44,4 +44,6 @@
         "X": {"length": 123869142, "synonyms": []},
         "MT": {"length": 16727, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "103",
 }

stdpopsim/catalog/CanFam/species.py

@@ -46,7 +46,7 @@
     "38": 1.4363726512881696e-08,
     "X": 9.506483722244087e-09,
     "MT": 0,
-    "Y": 0,  # manually set to 0 because it's not in the map
+    # "Y": 0,  # removed because not in the map or assembly
 }
 
 # Generic and chromosome-specific ploidy
@@ -92,9 +92,16 @@
     "38": _species_ploidy,
     "X": _species_ploidy,
     "MT": 1,
-    "Y": 1,
+    # "Y": 1, removed because it's not in the map
 }
 
+_mutation_rate = 4e-9
+_mutation_rate_data = {str(i): _mutation_rate for i in range(1, 39)}
+_mutation_rate_data["MT"] = (
+    _mutation_rate  # note this is likely incorrect but consistent with current setup
+)
+_mutation_rate_data["X"] = _mutation_rate
+
 _LindbladTohEtAl = stdpopsim.Citation(
     # Genome sequence, comparative analysis and haplotype structure of the
     # domestic dog.
@@ -139,10 +146,11 @@
         )
     )
 
-_genome = stdpopsim.Genome(
-    chromosomes=_chromosomes,
-    assembly_name=genome_data.data["assembly_name"],
-    assembly_accession=genome_data.data["assembly_accession"],
+_genome = stdpopsim.Genome.from_data(
+    genome_data=genome_data.data,
+    recombination_rate=_recombination_rate_data,
+    mutation_rate=_mutation_rate_data,
+    ploidy=_ploidy,
     citations=[
         _SkoglundEtAl.because(stdpopsim.CiteReason.MUT_RATE),
         _FrantzEtAl.because(stdpopsim.CiteReason.MUT_RATE),

stdpopsim/catalog/ChlRei/genome_data.py

@@ -21,4 +21,6 @@
         "16": {"length": 7783580, "synonyms": []},
         "17": {"length": 7188315, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/DroMel/genome_data.py

@@ -12,4 +12,6 @@
         "Y": {"length": 3667352, "synonyms": []},
         "mitochondrion_genome": {"length": 19524, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/DroSec/genome_data.py

@@ -10,4 +10,6 @@
         "X": {"length": 22909512, "synonyms": []},
         "4": {"length": 1277805, "synonyms": []},
     },
+    "assembly_source": "manual",
+    "assembly_build_version": None,
 }

stdpopsim/catalog/EscCol/genome_data.py

@@ -3,4 +3,6 @@
     "assembly_accession": "GCA_000005845.2",
     "assembly_name": "ASM584v2",
     "chromosomes": {"Chromosome": {"length": 4641652, "synonyms": []}},
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/EscCol/species.py

@@ -29,7 +29,6 @@
 )
 
 _species_ploidy = 1
-
 _chromosomes = []
 for name, data in genome_data.data["chromosomes"].items():
     _chromosomes.append(
@@ -46,11 +45,24 @@
         )
     )
 
-_genome = stdpopsim.Genome(
-    chromosomes=_chromosomes,
+
+_ploidy_data = {str(i.id): _species_ploidy for i in _chromosomes}
+
+_mutation_rate = 8.9e-11
+_mutation_rate_data = {str(i.id): _mutation_rate for i in _chromosomes}
+
+_recombination_rate = 8.9e-11
+_recombination_rate_data = {str(i.id): _recombination_rate for i in _chromosomes}
+
+_gene_conversion_length = 542
+_gene_conversion_data = {str(i.id): _gene_conversion_length for i in _chromosomes}
+_genome = stdpopsim.Genome.from_data(
+    genome_data=genome_data.data,
+    mutation_rate=_mutation_rate_data,
+    recombination_rate=_recombination_rate_data,
+    gene_conversion_length=_gene_conversion_data,
     bacterial_recombination=True,
-    assembly_name=genome_data.data["assembly_name"],
-    assembly_accession=genome_data.data["assembly_accession"],
+    ploidy=_ploidy_data,
     citations=[
         _wielgoss_et_al.because(
             {stdpopsim.CiteReason.MUT_RATE, stdpopsim.CiteReason.GENE_CONVERSION}

stdpopsim/catalog/GasAcu/genome_data.py

@@ -27,4 +27,6 @@
         "Y": {"length": 15859692, "synonyms": []},
         "MT": {"length": 16543, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "103",
 }

stdpopsim/catalog/HelAnn/genome_data.py

@@ -21,4 +21,6 @@
         "16": {"length": 206736614, "synonyms": []},
         "17": {"length": 195042445, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/HelMel/genome_data.py

@@ -34,4 +34,6 @@
         "20": {"length": 14871695, "synonyms": []},
         "21": {"length": 13359691, "synonyms": []},
     },
+    "assembly_source": "manual",
+    "assembly_build_version": None,
 }

stdpopsim/catalog/HomSap/genome_data.py

@@ -29,4 +29,6 @@
         "Y": {"length": 57227415, "synonyms": ["chrY"]},
         "MT": {"length": 16569, "synonyms": ["chrM"]},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/MusMus/genome_data.py

@@ -26,4 +26,6 @@
         "Y": {"length": 91455967, "synonyms": ["chrY"]},
         "MT": {"length": 16299, "synonyms": ["chrM"]},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/OrySat/genome_data.py

@@ -18,4 +18,6 @@
         "11": {"length": 29021106, "synonyms": []},
         "12": {"length": 27531856, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/PanTro/genome_data.py

@@ -29,4 +29,6 @@
         "X": {"length": 155549662, "synonyms": ["chrX"]},
         "Y": {"length": 26350515, "synonyms": ["chrY"]},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "103",
 }

stdpopsim/catalog/PapAnu/genome_data.py

@@ -26,4 +26,6 @@
         "X": {"length": 142711496, "synonyms": []},
         "Y": {"length": 8309886, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }

stdpopsim/catalog/PhoSin/genome_data.py

@@ -26,4 +26,6 @@
         "21": {"length": 35802323, "synonyms": []},
         "X": {"length": 131664873, "synonyms": []},
     },
+    "assembly_source": "ensembl",
+    "assembly_build_version": "113",
 }