PEtab import: Priors aren't support -> raise (#384)

So far, parPE doesn't support priors.

Most of the PEtab import was written before priors were introduced. That's why unfortunately there wasn't any warning/error in place so far.

benchmark_collection/all.sh

@@ -13,7 +13,6 @@ Borghans_BiophysChem1997
 Crauste_CellSystems2017
 Elowitz_Nature2000
 Fujita_SciSignal2010
-Schwen_PONE2014
 Sneyd_PNAS2002
 Weber_BMC2015
 Zheng_PNAS2012
@@ -47,7 +46,9 @@ Zheng_PNAS2012
 # Brannmark_JBC2010 FAILED: Expected llh 283.778227541074, got nllh 141.889034
 # Lucarelli_CellSystems2018
 # Fiedler_BMC2016 FAILED: Expected llh -117.16780323362, got nllh 109391.496205
-
+#
+# Unsupported priors:
+#   Schwen_PONE2014
 for model_name in $expected_to_work; do
   printf '=%.0s' {1..20}
   printf %s "${model_name}"

python/parpe/hdf5_pe_input.py

@@ -93,9 +93,6 @@ def __init__(self,
         self.petab_problem: petab.Problem = petab_problem
         self.amici_model: amici.Model = amici_model
 
-        # ensure we have valid inputs
-        petab.lint_problem(self.petab_problem)
-
         # index for no reference/preequilibration condition
         self.NO_PREEQ_CONDITION_IDX: int = NO_PREEQ_CONDITION_IDX
 
@@ -105,6 +102,28 @@ def __init__(self,
         # hdf5 dataset compression
         self.compression = "gzip"
 
+        self._check_support()
+
+    def _check_support(self):
+        """
+        Check if the model is supported by parPE
+        """
+        # ensure we have valid inputs
+        petab.lint_problem(self.petab_problem)
+
+        # We can't handle priors yet
+        #  Initialization priors would work in principle when we sample
+        #  starting points using petab. However, the current parpe C++
+        #  resampling would ignore the initialization prior.
+        #  And they wouldn't be supported for hierarchical optimization.
+        parameter_df = self.petab_problem.parameter_df
+        for col_id in [
+            ptc.INITIALIZATION_PRIOR_TYPE, ptc.INITIALIZATION_PRIOR_PARAMETERS,
+            ptc.OBJECTIVE_PRIOR_TYPE, ptc.OBJECTIVE_PRIOR_PARAMETERS
+        ]:
+            if (col := parameter_df.get(col_id)) is not None and col.notna().any():
+                raise NotImplementedError("Priors are not supported yet.")
+
     def generate_file(self, hdf5_file_name: str) -> None:
         """
         Create the output file