update vcell_pipeline download for pydantic and improved test.

python-utils/tests/citation_test.py

@@ -6,9 +6,47 @@ def test_citation() -> None:
     citation: Optional[CitationInfo] = getCitation(pubmedid="35367415")
     assert citation is not None
     assert citation.author == "Nosbisch, Jamie L"
-    assert citation.title == "A kinetic model of phospholipase C-γ1 linking structure-based insights to dynamics of enzyme autoinhibition and activation"
+    assert citation.title == ("A kinetic model of phospholipase C-γ1 linking structure-based insights to dynamics of "
+                              "enzyme autoinhibition and activation")
     assert citation.year == 2022
     assert citation.journal == "J Biol Chem"
 
     suggestedProjectName = getSuggestedProjectName(bm_key="12345", pub_info=None, citation_info=citation)
     assert suggestedProjectName == "VCDB_12345_Nosbisch_JBiolChem_2022"
+
+
+def test_get_citation():
+    pubmdeid = "35367415"
+    expected_title = ("A kinetic model of phospholipase C-γ1 linking structure-based insights to dynamics of "
+                      "enzyme autoinhibition and activation")
+    expected_year = 2022
+    expected_journal = "J Biol Chem"
+    expected_author = "Nosbisch, Jamie L"
+    expected_abstract = ("Phospholipase C-γ1 (PLC-γ1) is a receptor-proximal enzyme that promotes signal transduction "
+                         "through PKC in mammalian cells. Because of the complexity of PLC-γ1 regulation, a two-state "
+                         "(inactive/active) model does not account for the intricacy of activation and inactivation "
+                         "steps at the plasma membrane. Here, we introduce a structure-based kinetic model of PLC-γ1, "
+                         "considering interactions of its regulatory Src homology 2 (SH2) domains and perturbation of "
+                         "those dynamics upon phosphorylation of Tyr783, a hallmark of activation. For PLC-γ1 "
+                         "phosphorylation to dramatically enhance enzyme activation as observed, we found that high "
+                         "intramolecular affinity of the C-terminal SH2 (cSH2) domain-pTyr783 interaction is critical, "
+                         "but this affinity need not outcompete the autoinhibitory interaction of the cSH2 domain. "
+                         "Under conditions for which steady-state PLC-γ1 activity is sensitive to the rate of Tyr783 "
+                         "phosphorylation, maintenance of the active state is surprisingly insensitive to the "
+                         "phosphorylation rate, since pTyr783 is well protected by the cSH2 domain while the enzyme is "
+                         "active. In contrast, maintenance of enzyme activity is sensitive to the rate of PLC-γ1 "
+                         "membrane (re)binding. Accordingly, we found that hypothetical PLC-γ1 mutations that either "
+                         "weaken autoinhibition or strengthen membrane binding influence the activation kinetics "
+                         "differently, which could inform the characterization of oncogenic variants. Finally, we used "
+                         "this newly informed kinetic scheme to refine a spatial model of PLC/PKC polarization during "
+                         "chemotaxis. The refined model showed improved stability of the polarized pattern while "
+                         "corroborating previous qualitative predictions. As demonstrated here for PLC-γ1, this "
+                         "approach may be adapted to model the dynamics of other receptor- and membrane-proximal "
+                         "enzymes.")
+    citation = getCitation(pubmedid=pubmdeid)
+    assert citation is not None
+    assert citation.abstract == expected_abstract
+    assert citation.author == expected_author
+    assert citation.title == expected_title
+    assert citation.year == expected_year
+    assert citation.journal == expected_journal

python-utils/vcutils/vcell_pipeline/datamodels.py

@@ -23,12 +23,12 @@ class Publication(BaseModel):
     title: str
     authors: list[str]
     year: int
-    citation: Optional[str]
-    pubmedid: Optional[str]
-    doi: Optional[str]
-    endnoteid: Optional[str]
-    url: Optional[str]
-    wittid: Optional[str]
+    citation: Optional[str] = None
+    pubmedid: Optional[str] = None
+    doi: Optional[str] = None
+    endnoteid: Optional[str] = None
+    url: Optional[str] = None
+    wittid: Optional[str] = None
     biomodelReferences: list[BiomodelReference]
     mathmodelReferences: list[MathmodelReference]
     date: str

python-utils/vcutils/vcell_pipeline/download_vcell_omex.py

@@ -5,8 +5,8 @@
 from pydantic import BaseModel
 
 from vcutils.common.api_utils import download_file
-from .citation import getCitation, CitationInfo, getSuggestedProjectName
-from .datamodels import Publication
+from vcutils.vcell_pipeline.citation import getCitation, CitationInfo, getSuggestedProjectName
+from vcutils.vcell_pipeline.datamodels import Publication
 
 
 class ExportStatus(BaseModel):
@@ -26,14 +26,14 @@ def write_log(entry: ExportStatus, log_path: Path) -> None:
 def download_published_omex(api_base_url: str, out_dir: Path) -> None:
     log_path = Path(out_dir, "export.log")
 
-    response = requests.get(f"{api_base_url}/publication", headers={'Accept': 'application/json'}, verify=False)
+    response = requests.get(f"{api_base_url}/publication", headers={'Accept': 'application/json'}, verify=True, timeout=600)
     publication_json = response.json()
     pubs = [Publication(**jsonDict) for jsonDict in publication_json]
 
     for pub in pubs:
         if len(pub.biomodelReferences) == 0:
             continue
-
+        print(f"Processing {pub.pubKey}, title: {pub.title}, year: {pub.year}, bimodels: {pub.biomodelReferences}")
         bmKey = pub.biomodelReferences[0].bmKey
         pubmedId: Optional[str] = pub.pubmedid
         citationInfo: Optional[CitationInfo] = None