validation of gw happens inside validate_input_config_mtm

src/cryo_challenge/_commands/run_map2map_pipeline.py

@@ -8,7 +8,6 @@
 
 from .._map_to_map.map_to_map_pipeline import run
 from ..data._validation.config_validators import validate_input_config_mtm
-from ..data._validation.config_validators import GWConfig
 
 
 def add_args(parser):
@@ -37,11 +36,6 @@ def main(args):
         config = yaml.safe_load(file)
 
     validate_input_config_mtm(config)
-    if "gromov_wasserstein_extra_params" in config["analysis"]:
-        _ = GWConfig(
-            **config["analysis"]["gromov_wasserstein_extra_params"]
-        ).model_dump()
-
     warnexists(config["output"])
     mkbasedir(os.path.dirname(config["output"]))
 

src/cryo_challenge/data/_validation/config_validators.py

@@ -146,6 +146,44 @@ def validate_config_mtm_analysis_normalize(config_analysis_normalize: dict) -> N
     return
 
 
+class GWConfig(BaseModel):
+    """
+    Gromov-Wasserstein extra parameters config
+
+    Parameters
+    ----------
+    top_k : int
+        Number of voxels to use (ranked according to highest mass)
+    n_downsample_pix : int
+        Number of pixels to downsample to (in each dimension)
+    exponent : float
+        exponential weighting of GW cost. Base cost is Euclindean distance
+    cost_scale_factor : float
+        multiplicative scaling factor for the cost (before exponentiation)
+    element_wise : bool
+        dask parralelization: whether to call dask.compute on each map-to-map computation, or naively loop through each (80) submitted maps row-wise
+    slurm : bool
+        parallelization configuration: whether to use dask_hpc_runner.SlurmRunner as a runner for dask.Client(runner)
+    scheduler : Optional[str]
+        string argument to dask.compute
+    local_directory : Optional[str]
+        directory for dask.distributed.Client
+    """
+
+    top_k: int
+    n_downsample_pix: int
+    exponent: float
+    cost_scale_factor: float
+    element_wise: bool
+    slurm: bool
+    scheduler: Optional[str] = None
+    local_directory: Optional[str] = None
+
+
+def validate_config_mtm_analysis_gromov_wasserstein_extra_arams(config):
+    _ = GWConfig(**config).model_dump()
+
+
 def validate_config_mtm_analysis(config_analysis: dict) -> None:
     """
     Validate the analysis part of the config dictionary for the MapToMap config.
@@ -167,6 +205,14 @@ def validate_config_mtm_analysis(config_analysis: dict) -> None:
 
     validate_generic_config(config_analysis, keys_and_types)
     validate_config_mtm_analysis_normalize(config_analysis["normalize"])
+    if "gromov_wasserstein_extra_params" in config_analysis:
+        validate_config_mtm_analysis_gromov_wasserstein_extra_arams(
+            config_analysis["gromov_wasserstein_extra_params"]
+        )
+    else:
+        Warning(
+            "No Gromov-Wasserstein extra parameters found in config, so not validated."
+        )
     return
 
 
@@ -375,37 +421,3 @@ def check_voxel_size(cls, value):
         if value <= 0:
             raise ValueError("Voxel size must be positive.")
         return value
-
-
-class GWConfig(BaseModel):
-    """
-    Config for the GW distance computation
-
-    Parameters
-    ----------
-    top_k : int
-        Number of voxels to use (ranked according to highest mass)
-    n_downsample_pix : int
-        Number of pixels to downsample to (in each dimension)
-    exponent : float
-        exponential weighting of GW cost. Base cost is Euclindean distance
-    cost_scale_factor : float
-        multiplicative scaling factor for the cost (before exponentiation)
-    element_wise : bool
-        dask parralelization: whether to call dask.compute on each map-to-map computation, or naively loop through each (80) submitted maps row-wise
-    slurm : bool
-        parallelization configuration: whether to use dask_hpc_runner.SlurmRunner as a runner for dask.Client(runner)
-    scheduler : Optional[str]
-        string argument to dask.compute
-    local_directory : Optional[str]
-        directory for dask.distributed.Client
-    """
-
-    top_k: int
-    n_downsample_pix: int
-    exponent: float
-    cost_scale_factor: float
-    element_wise: bool
-    slurm: bool
-    scheduler: Optional[str] = None
-    local_directory: Optional[str] = None