adjusted migration

Initial bayer app

Show Pack Classification

Adjusted docker compose to bayer specifics

Adjusted Dockerfile for Bayer

Adding secret flags to group, add secret pools to packages

Adjusted View for Package creation

Prep configs, added Package Create Modal

wip

More on PES

wip

wip

Wip

minor

PW interactions

API PES

wip

Make Select Widget reflect required

make required generallay available

Update UI if pathway mode is set to build

Added ais

circle adjustments

Initial Zoom, fix AD Creation

wip

auth log, bb4g fix

missing import

Added viz hint if PES is part of reaction

Add Edge check for pes

flip boolean

...

pes

Added extra

...

In / Out Edges Viz, Submitting Button Text

...

Make PES Link clickable

Return proper http response instead of error

Fixed error return, removed unused options

Fix PES Link HTML for other entities

Fixed molfile assignment, adjusted Export

Package Export/Import cycle

highlight Description links

implemented non persistent

Harmonised proposed field in Json output

Added pesLink field to PW Api output

PES Fields in API Output

removed debug

Fix Classification import, Fix PES Deserialization

underline pes link in templates

Fix alter name/desc for node, make /node /edge funcitonal

provide setting link and copy button

Implemented Compound Names / Reaction Names View Option

Unconnected Nodes

Make links thicker, reduce timeout trigger time

Show proposed info in popover

Pathway Build no stereo removal

Include probs in reaction name option viz

Detect clicks outside nodes/edges

Provide proper Error Pages

View Package Perm

wip

sync

Auth log

auth log leftovers

...

secret packs viz

auth log for api

model stats

Fix Package Adjustment

Adjust Group Auth Log

Fix Secret image size in Navbar

leftover

...

minor

epdb/models.py

@@ -208,6 +208,7 @@ class Group(TimeStampedModel):
     name = models.TextField(blank=False, null=False, verbose_name="Group name")
     owner = models.ForeignKey("User", verbose_name="Group Owner", on_delete=models.CASCADE)
     public = models.BooleanField(verbose_name="Public Group", default=False)
+    secret = models.BooleanField(verbose_name="Secret Group", default=False)
     description = models.TextField(
         blank=False, null=False, verbose_name="Descriptions", default="no description"
     )
@@ -579,6 +580,10 @@ class ReactionIdentifierMixin(ExternalIdentifierMixin):
     def get_uniprot_identifiers(self):
         return self.get_external_identifier("UniProt")
 
+    def contains_pes(self):
+        from bayer.models import PESStructure
+        return any([isinstance(o, PESStructure) for o in self.educts.all()]) or any(
+            [isinstance(o, PESStructure) for o in self.products.all()])
 
 ##############
 # EP Objects #
@@ -900,7 +905,11 @@ class Compound(
 
         standardized_smiles = FormatConverter.standardize(smiles, remove_stereo=True)
 
+        subclasses = CompoundStructure.__subclasses__()
+
         qs = CompoundStructure.objects.filter(smiles=smiles, compound__package=package)
+        if subclasses:
+            qs = qs.not_instance_of(*subclasses)
 
         # Check if we find a direct match for a given SMILES
         if qs.exists():
@@ -914,6 +923,8 @@ class Compound(
             return found_compound
 
         qs = CompoundStructure.objects.filter(smiles=standardized_smiles, compound__package=package)
+        if subclasses:
+            qs = qs.not_instance_of(*subclasses)
 
         # Check if we can find the standardized one
         if qs.exists():
@@ -1253,7 +1264,7 @@ class CompoundStructure(
             raise ValueError("Unpersisted Compound! Persist compound first!")
 
         cs = CompoundStructure()
-        # Clean for potential XSS
+
         if name is not None:
             cs.name = name
 
@@ -2804,6 +2815,58 @@ class PackageBasedModel(EPModel):
     )
     multigen_eval = models.BooleanField(null=False, blank=False, default=False)
 
+    def parameters(self):
+        params = {
+            "Model Evaluation Threshold": f"{self.threshold:.2f}",
+            "Multi Gen Evaluation": "Yes" if self.multigen_eval else "No",
+        }
+
+        if self.app_domain:
+            params["Applicability Domain Num Neighbors"] = f"{self.app_domain.num_neighbours:.2f}"
+            params["Applicability Domain Reliability Threshold"] = (
+                f"{self.app_domain.reliability_threshold:.2f}"
+            )
+            params["Applicability Domain Local Compatibility Threshold"] = (
+                f"{self.app_domain.local_compatibilty_threshold:.2f}"
+            )
+
+        return params
+
+    def statistics(self):
+        from sklearn.metrics import auc
+
+        recall = list(self.eval_results["average_recall_per_threshold"].values())
+        precision = list(self.eval_results["average_precision_per_threshold"].values())
+        mg_recall = list(
+            self.eval_results.get("multigen_average_recall_per_threshold", {}).values()
+        )
+        mg_precision = list(
+            self.eval_results.get("multigen_average_precision_per_threshold", {}).values()
+        )
+
+        return {
+            "accuracy": [
+                self.eval_results["average_accuracy"],
+                self.eval_results.get("multigen_average_accuracy"),
+            ],
+            "precision": [
+                self.eval_results["average_precision_per_threshold"][f"{self.threshold:.2f}"],
+                self.eval_results.get("multigen_average_precision_per_threshold", {}).get(
+                    f"{self.threshold:.2f}"
+                ),
+            ],
+            "recall": [
+                self.eval_results["average_recall_per_threshold"][f"{self.threshold:.2f}"],
+                self.eval_results.get("multigen_average_recall_per_threshold", {}).get(
+                    f"{self.threshold:.2f}"
+                ),
+            ],
+            "Area under PR Curve": [
+                auc(recall, precision),
+                auc(mg_recall, mg_precision) if self.multigen_eval else None,
+            ],
+        }
+
     @property
     def pr_curve(self):
         if self.model_status != self.FINISHED:
@@ -3027,8 +3090,6 @@ class PackageBasedModel(EPModel):
                 thresholds.append(np.float64(threshold))
                 thresholds.sort()
 
-            logger.info(f"Thresholds: {thresholds}")
-
             precision = {f"{t:.2f}": [] for t in thresholds}
             recall = {f"{t:.2f}": [] for t in thresholds}
 
@@ -3077,14 +3138,18 @@ class PackageBasedModel(EPModel):
             for t in thresholds:
                 for true, pred in zip(pathways, pred_pathways):
                     acc, pre, rec = multigen_eval(true, pred, t)
-                    if abs(t - threshold) < 0.01:
-                        mg_acc = acc
+
+                    if f"{t:.2f}" == f"{threshold:.2f}":
+                        mg_acc += acc
+
                     precision[f"{t:.2f}"].append(pre)
                     recall[f"{t:.2f}"].append(rec)
 
+            avg_mg_acc = mg_acc / len(root_compounds)
             precision = {k: sum(v) / len(v) if len(v) > 0 else 0 for k, v in precision.items()}
             recall = {k: sum(v) / len(v) if len(v) > 0 else 0 for k, v in recall.items()}
-            return mg_acc, precision, recall
+            logger.info("Average Multigen Accuracy: {:.2f}".format(avg_mg_acc))
+            return avg_mg_acc, precision, recall
 
         # If there are eval packages perform single generation evaluation on them instead of random splits
         if self.eval_packages.count() > 0: