From c8bdd1a71dbeb35242fe0375f618c356b3bf6e8d Mon Sep 17 00:00:00 2001
From: christinehc <christine.chang@pnnl.gov>
Date: Wed, 29 Jan 2025 10:48:10 -0800
Subject: [PATCH] [in-prog]fix: attempt chem file regeneration and linkml
 update

---
 build_script.py                               |  1 +
 sampleChemMapping/map_samples_to_chemicals.py |  2 +-
 srpAnalytics.yaml                             | 48 +++++++++++--------
 3 files changed, 31 insertions(+), 20 deletions(-)

diff --git a/build_script.py b/build_script.py
index 92b0dd6..27cace8 100644
--- a/build_script.py
+++ b/build_script.py
@@ -77,6 +77,7 @@ def runSampMap(
     """
     run sample mapping
     """
+    print("drcfiles:", drcfiles)
     drc = ",".join(drcfiles)
     args = (
         f" --sample_id={sid}"
diff --git a/sampleChemMapping/map_samples_to_chemicals.py b/sampleChemMapping/map_samples_to_chemicals.py
index 2afd2b4..7ba13aa 100644
--- a/sampleChemMapping/map_samples_to_chemicals.py
+++ b/sampleChemMapping/map_samples_to_chemicals.py
@@ -604,7 +604,7 @@ def main():
             )
 
         if args.is_chem:
-            nas = bmds["Chemical_ID"].isna()
+            nas = bmds[bmds["Chemical_ID"].isna()]["Chemical_ID"]
             to_remove = set(nas) - set(chem_sample["chemical_id"])
             bmds = bmds[~bmds["Chemical_ID"].isin(to_remove)]
             curves = curves[~curves["Chemical_ID"].isin(to_remove)]
diff --git a/srpAnalytics.yaml b/srpAnalytics.yaml
index 4bdc059..eaa4f46 100644
--- a/srpAnalytics.yaml
+++ b/srpAnalytics.yaml
@@ -15,11 +15,21 @@ slots:
     range: linkml:Any
     identifier: true
     slot_uri: schema:identifier
+  chemical_id:
+    description: Unique identifier for every chemical in the database
+    range: linkml:Any
+    identifier: true
+    slot_uri: schema:identifier
   Sample_ID:
     description: Unique identifier for every sample in the database
     range: linkml:Any
     identifier: true
     slot_uri: schema:identifier
+  sample_id:
+    description: Unique identifier for every sample in the database
+    range: linkml:Any
+    identifier: true
+    slot_uri: schema:identifier
   cas_number:
     description: CAS identifier for a specific chemical
   concentration:
@@ -29,52 +39,52 @@ slots:
     description: Unit of concentration
   Project:
     description: name of project contributing the dataset
-  Model:
+  model:
     description: Model with best fit to data
-  BMD10:
+  bmd10:
     description: Dose at which 10% of fish are affected
     range: linkml:Any
-  BMD50:
+  bmd50:
     description: Dose at which 50% of fish are affected
     range: linkml:Any
-  Min_Dose:
+  min_dose:
     description: Minimum dose administered
     range: linkml:Any
-  Max_Dose:
+  max_dose:
     description: Maximum dose administered
     range: linkml:Any    
-  AUC_Norm:
+  auc_norm:
     description: normalized area under the curve
     range: linkml:Any
-  BMD_Analysis_Flag:
+  bmd_analysis_flag:
     description: quality of curve fit
     range: integer
-  IncludeInPortal:
+  include_in_portal:
     description:
-  End_Point_Name:
+  end_point_name:
     description:
-  Description:
+  description:
     description: Name of endpoint
-  endPointLink:
+  end_point_link:
     description: Link to endpoint description
-  DataQC_Flag:
+  data_qc_flag:
     description: Quality of Data
-  Dose:
+  dose:
     description: Dose of chemical
     range: float
-  Response:
+  response:
     description: Response in terms of percentage of fish alive
     range: float
-  CI_Lo:
+  ci_lo:
     description: Low range of confidence interval
     range: float
-  CI_Hi:
+  ci_hi:
     description: High range of confidence interval
     range: float
-  X_vals:
+  x_vals:
     description: X coordinates of fitted curve
     range: linkml:Any
-  Y_vals:
+  y_vals:
     description: Y coordinates of fitted curve
     range: linkml:Any
 
@@ -193,7 +203,7 @@ classes:
       - end_point_name
       - description
       - end_point_link
-      - data_qc_Flag
+      - data_qc_flag
   zebrafishSampDoseResponse:
     description: Dose response datapoints of sample extracts in zebrafish
     slots: