Add provenance and URL

multitask/etl/etl_baumgartner_cn.py

@@ -33,6 +33,7 @@
 
 """
 
+import datetime
 from ord_schema.proto.reaction_pb2 import *
 from ord_schema.proto.dataset_pb2 import *
 from ord_schema import validations
@@ -126,6 +127,9 @@ def inner_loop(row: pd.Series, stock_df: pd.DataFrame) -> Reaction:
     # Specify reaction outcome
     specify_outcome(reaction, row)
 
+    # Add provenance
+    add_provenance(reaction)
+
     return reaction
 
 
@@ -604,6 +608,7 @@ def add_standard(measurement: ProductMeasurement, row: pd.Series):
 def define_measurement(measurement: ProductMeasurement, row: pd.Series):
     measurement.analysis_key = "LCMS"
     measurement.type = ProductMeasurement.YIELD
+    measurement.uses_internal_standard = True
     rxn_yield = row["Reaction Yield"]
 
     try:
@@ -613,27 +618,10 @@ def define_measurement(measurement: ProductMeasurement, row: pd.Series):
         rxn_yield = rxn_yield / 100
     except:
         pass
-    # try:
-    #     rxn_yield = rxn_yield.replace("≥", "")
-    # except:
-    #     pass
-    # try:
-    #     rxn_yield = rxn_yield.replace("%", "")
-    # except:
-    #     pass
-    # try:
-    #     rxn_yield = float(rxn_yield)
-    # except:
-    #     pass
+
     if rxn_yield < 200:
         measurement.percentage.value = rxn_yield * 100
 
-    # measurement.retention_time.value = row[
-    #    "2-Fluoro-3,3'-bipyridine Retention time in min"
-    # ]
-    # measurement.retention_time.units = Time.MINUTE
-
-
 # nucleophile ID must be given. Since electrophile stays constant, the product
 # can be inferred from the nucleophile alone
 def specify_outcome_details(nuc_id: str):
@@ -680,12 +668,17 @@ def add_provenance(reaction: Reaction):
     provenance = reaction.provenance
     provenance.doi = "10.1021/acs.oprd.9b00236"
     provenance.publication_url = "http://doi.org/10.1021/acs.oprd.9b00236"
-    creator = provenance.record_created.person
-    creator.username = "dswigh"
-    creator.name = "Daniel Wigh"
-    creator.orcid = "0000-0002-0494-643X"
-    creator.organization = "University of Cambridge"
-    creator.email = "[email protected]"
+    event = RecordEvent(
+        time={"value": str(datetime.datetime.now())},
+        person={
+            "username": "dswigh",
+            "name": "Daniel Wigh",
+            "orcid": "0000-0002-0494-643X",
+            "organization": "University of Cambridge",
+            "email": "[email protected]"
+        },
+    )
+    provenance.record_created.CopyFrom(event)
 
 
 if __name__ == "__main__":

multitask/etl/etl_baumgartner_suzuki.py

@@ -5,6 +5,7 @@
 
 """
 
+import datetime
 from ord_schema.proto.reaction_pb2 import *
 from ord_schema.proto.dataset_pb2 import *
 from ord_schema.message_helpers import find_submessages
@@ -81,6 +82,9 @@ def inner_loop(row: pd.Series) -> Reaction:
     # Specify reaction outcome
     specify_outcome(reaction, row)
 
+    # Add provenance
+    add_provenance(reaction)
+
     return reaction
 
 
@@ -424,6 +428,7 @@ def define_measurement(measurement: ProductMeasurement, row: pd.Series):
         "2-Fluoro-3,3'-bipyridine Retention time in min"
     ]
     measurement.retention_time.units = Time.MINUTE
+    measurement.uses_internal_standard = True
 
 
 def specify_outcome(reaction: Reaction, row: pd.Series):
@@ -461,12 +466,18 @@ def add_provenance(reaction: Reaction):
     provenance = reaction.provenance
     provenance.doi = "10.1039/c8re00032h"
     provenance.publication_url = "http://doi.org/10.1039/c8re00032h"
-    creator = provenance.record_created.person
-    creator.username = "marcosfelt"
-    creator.name = "Kobi Felton"
-    creator.orcid = "0000-0002-3616-4766"
-    creator.organization = "University of Cambridge"
-    creator.email = "[email protected]"
+    event = RecordEvent(
+        time={"value": str(datetime.datetime.now())},
+        person={
+            "username": "marcosfelt",
+            "name": "Kobi Felton",
+            "orcid": "0000-0002-3616-4766",
+            "organization": "University of Cambridge",
+            "email": "[email protected]"
+
+        },
+    )
+    provenance.record_created.CopyFrom(event)
 
 
 if __name__ == "__main__":

multitask/etl/etl_reizman_suzuki.py

@@ -5,6 +5,7 @@
 
 """
 
+import datetime
 from multitask.utils import *
 from ord_schema.proto.reaction_pb2 import *
 from ord_schema.proto.dataset_pb2 import *
@@ -80,6 +81,9 @@ def inner_loop(row: pd.Series, case) -> Reaction:
     # Specify reaction outcome
     specify_outcome(case, reaction, row)
 
+    # Add provenance
+    add_provenance(reaction)
+
     return reaction
 
 
@@ -380,6 +384,7 @@ def define_measurement(measurement: ProductMeasurement, row: pd.Series):
     measurement.analysis_key = "LCMS"
     measurement.type = ProductMeasurement.YIELD
     measurement.percentage.value = row["yld"]
+    measurement.uses_internal_standard = True
 
 
 products_by_case = {
@@ -431,14 +436,20 @@ def specify_outcome(case, reaction: Reaction, row: pd.Series):
 
 def add_provenance(reaction: Reaction):
     provenance = reaction.provenance
-    provenance.doi = "110.1039/C6RE00153J"
+    provenance.doi = "10.1039/C6RE00153J"
     provenance.publication_url = "http://doi.org/10.1039/C6RE00153J"
-    creator = provenance.record_created.person
-    creator.username = "marcosfelt"
-    creator.name = "Kobi Felton"
-    creator.orcid = "0000-0002-3616-4766"
-    creator.organization = "University of Cambridge"
-    creator.email = "[email protected]"
+    event = RecordEvent(
+        time={"value": str(datetime.datetime.now())},
+        person={
+            "username": "marcosfelt",
+            "name": "Kobi Felton",
+            "orcid": "0000-0002-3616-4766",
+            "organization": "University of Cambridge",
+            "email": "[email protected]"
+
+        },
+    )
+    provenance.record_created.CopyFrom(event)
 
 
 if __name__ == "__main__":