Merge pull request #4 from z-haiman/develop

Develop 0.3.0
z-haiman · Jan 19, 2024 · 3e3261a · 3e3261a
2 parents fb38c0d + 9d223e6
commit 3e3261a
Show file tree

Hide file tree

Showing 40 changed files with 45,615 additions and 35,239 deletions.
diff --git a/.deprecated/code/curation/RBC-GEM_v0.2.0_v0.3.0.ipynb b/.deprecated/code/curation/RBC-GEM_v0.2.0_v0.3.0.ipynb
diff --git a/code/notebooks/annotation/AnnotateModel_RBC-GEM.ipynb b/code/notebooks/annotation/AnnotateModel_RBC-GEM.ipynb
diff --git a/code/notebooks/annotation/ExtractAnnotations_MIM.ipynb b/code/notebooks/annotation/ExtractAnnotations_MIM.ipynb
diff --git a/code/notebooks/annotation/ExtractAnnotations_MetAtlas.ipynb b/code/notebooks/annotation/ExtractAnnotations_MetAtlas.ipynb
diff --git a/code/notebooks/annotation/ExtractAnnotations_RBC-GEM.ipynb b/code/notebooks/annotation/ExtractAnnotations_RBC-GEM.ipynb
diff --git a/code/notebooks/annotation/ExtractAnnotations_UniProt.ipynb b/code/notebooks/annotation/ExtractAnnotations_UniProt.ipynb
diff --git a/code/notebooks/annotation/ReconcileAnnotations_RBC-GEM.ipynb b/code/notebooks/annotation/ReconcileAnnotations_RBC-GEM.ipynb
diff --git a/code/notebooks/data/AggregateData_Proteomics.ipynb b/code/notebooks/data/AggregateData_Proteomics.ipynb
diff --git a/code/notebooks/data/EvidenceGenes_RBC-GEM.ipynb b/code/notebooks/data/EvidenceGenes_RBC-GEM.ipynb
diff --git a/code/notebooks/data/IdentifierUpdates_Proteomics.ipynb b/code/notebooks/data/IdentifierUpdates_Proteomics.ipynb
diff --git a/code/src/rbc_gem_utils/__init__.py b/code/src/rbc_gem_utils/__init__.py
@@ -21,7 +21,9 @@
     ANNOTATION_PATH,
     INTERIM_PATH,
     EXTERNAL_PATH,
+    CURATION_PATH,
     GEM_NAME, 
+    GEM_URL,
     COBRA_CONFIGURATION,
     build_string,
     split_string, 

diff --git a/code/src/rbc_gem_utils/annotation.py b/code/src/rbc_gem_utils/annotation.py
@@ -1,7 +1,51 @@
 import pandas as pd
 from cobra import DictList
+import logging
+from rbc_gem_utils.util import ensure_iterable, build_string, check_if_valid
+
+LOGGER = logging.getLogger(__name__)
+
+def set_sbo_default_annotations(model, annotation_type, verbose=False):
+    """Set default SBO annotations on the model."""
+    annotation_type = check_if_valid(annotation_type, ["reactions", "metabolites", "genes"], "Unrecognized annotations for Human-GEM")
+    for attr in annotation_type:
+        for model_object in getattr(model, attr):
+            if attr == "reactions":
+                if model_object.boundary and model_object in model.exchanges:
+                    sbo_term = "SBO:0000627" # Exchange reaction
+                elif model_object.boundary and model_object in model.demands:
+                    sbo_term = "SBO:0000628" # Demand reaction
+                elif model_object.boundary and model_object in model.sinks:
+                    sbo_term = "SBO:0000632" # Sink reaction
+                elif len(model_object.compartments) > 1:
+                    sbo_term = "SBO:0000185" # Transport reaction
+                else:
+                    sbo_term = "SBO:0000176" # Biochemical reaction
+            elif attr == "metabolites":
+                sbo_term = "SBO:0000247" # Metabolite
+            elif attr == "genes":
+                sbo_term = "SBO:0000243" # Gene
+            else:
+                continue
+
+            try:
+                assert model_object.annotation["sbo"] == sbo_term
+            except KeyError:
+                msg = f"SBO term set for {model_object.id}"
+                LOGGER.info(msg)
+                if verbose:
+                    print(msg)
+                model_object.annotation["sbo"] = sbo_term
+            except AssertionError:
+                msg = f"SBO term changed for {model_object.id}: {model_object.annotation['sbo']} --> {sbo_term}"
+                LOGGER.info(msg)
+                if verbose:
+                    print(msg)
+                model_object.annotation["sbo"] = sbo_term
+
+    return model
+
 
-from rbc_gem_utils.util import ensure_iterable, build_string
 
 def get_id_annotation_mapping(objects, annotation_key, default=None):
     """Return a dictionary containing model identifiers mapped to annotation values.

diff --git a/code/src/rbc_gem_utils/database/metatlas.py b/code/src/rbc_gem_utils/database/metatlas.py
@@ -107,7 +107,7 @@ def get_annotations_HumanGEM(
     return df_annotations
 
 
-def download_database_HumanGEM(annotation_type=None, database_dirpath=None, model_filetype=None, model_version=None):
+def download_database_HumanGEM(annotation_type=None, database_dirpath=None, model_filetype=None, model_version="xml"):
     """Download the HumanGEM database files. Requires internet connection.
 
     Default values are used based on the RBC-GEM repository format.
@@ -140,10 +140,7 @@ def download_database_HumanGEM(annotation_type=None, database_dirpath=None, mode
     else:
         annotation_type = check_if_valid(annotation_type, valid, "Unrecognized annotations for Human-GEM")
 
-    if model_filetype is None:
-        model_filetype = ["xml"]
-    else:
-        model_filetype = check_if_valid(model_filetype, HUMANGEM_MODEL_FILETYPES, "Unrecognized filetypes for Human-GEM")
+    model_filetype = check_if_valid(model_filetype, HUMANGEM_MODEL_FILETYPES, "Unrecognized filetypes for Human-GEM")
 
     if database_dirpath is not None:
         # Ensure the path exists

diff --git a/code/src/rbc_gem_utils/util.py b/code/src/rbc_gem_utils/util.py
@@ -22,10 +22,13 @@
 PROCESSED_PATH = "/data/processed"
 EXTERNAL_PATH = "/data/external"
 ANNOTATION_PATH = "/data/annotation"
+CURATION_PATH = "/data/curation"
 DATABASE_PATH = f"{EXTERNAL_PATH}/database"
 
 
 GEM_NAME = "RBC-GEM"
+GEM_URL = f"{RAW_GH_URL}/z-haiman/{GEM_NAME}"
+GEM_MODEL_FILETYPES = {'mat', 'json', 'xml', 'yml'}
 
 
 def show_versions():

diff --git a/data/annotation/genes_MIM.tsv b/data/annotation/genes_MIM.tsv