Fix remaining test issues

Author: RalfG

im2deep/_io_helpers.py

@@ -18,6 +18,7 @@
 import logging
 from copy import deepcopy
 from pathlib import Path
+from typing import Any
 
 import click
 import numpy as np
@@ -28,13 +29,25 @@
 from rich.console import Console
 from rich.logging import RichHandler
 
-from im2deep.exceptions import IM2DeepError
+from im2deep.exceptions import IM2DeepError, PSMMetadataError
 from im2deep.utils import ccs2im, im2ccs
 
 console = Console()
 
 LOGGER = logging.getLogger(__name__)
 
+ConvertibleToFloat = str | float | int | np.floating[Any] | np.ndarray[Any, Any]
+
+
+def _normalize_ccs_metadata_value(value: ConvertibleToFloat) -> str:
+    """Validate and normalize a CCS metadata value to the canonical string form."""
+    try:
+        return str(float(value))
+    except (TypeError, ValueError) as exc:
+        raise PSMMetadataError(
+            f"Invalid CCS metadata value {value!r}; expected a numeric string or number."
+        ) from exc
+
 
 def parse_input(
     input_file: str | Path | PSMList | pd.DataFrame, filetype: str | None = None
@@ -86,7 +99,7 @@ def parse_input(
                 if "CCS" in row:
                     if precursor.metadata is None:
                         precursor.metadata = {}
-                    precursor.metadata["CCS"] = float(row["CCS"])  # type: ignore
+                    precursor.metadata["CCS"] = _normalize_ccs_metadata_value(row["CCS"])  # type: ignore
                 list_of_precursors.append(precursor)
             except Exception as e:
                 LOGGER.warning("Error parsing row %d: %s. Skipping.", idx, e)
@@ -195,7 +208,7 @@ def _parse_legacy_format(input_file: str | Path) -> PSMList:
                 charge=int(row["charge"]),
             )
             if has_ccs:
-                metadata = {"CCS": float(row["CCS"])}
+                metadata = {"CCS": _normalize_ccs_metadata_value(row["CCS"])}
 
             LOGGER.debug(f"Parsed PSM: {peptidoform} with metadata: {metadata}")
             precursor = PSM(peptidoform=peptidoform, metadata=metadata, spectrum_id=idx)
@@ -263,19 +276,22 @@ def validate_psm_list(psm_list: PSMList, needs_target: bool = False) -> PSMList:
         )
 
         # TODO: Could be vectorized over all ion mobility values
-        # If ion_mobility is present, convert to CCS
+        # Normalize targets while preserving the PSM metadata str -> str contract.
         for psm in psm_list_filtered:
+            if psm.metadata is None:
+                psm.metadata = {}
+
             if psm.ion_mobility is not None:
-                if psm.metadata is None:
-                    psm.metadata = {}
                 if "CCS" not in psm.metadata:
-                    psm.metadata["CCS"] = str(
+                    psm.metadata["CCS"] = _normalize_ccs_metadata_value(
                         im2ccs(
                             psm.ion_mobility,
                             psm.peptidoform.theoretical_mz,
                             psm.peptidoform.precursor_charge,
                         )
                     )
+            elif psm.metadata.get("CCS") is not None:
+                psm.metadata["CCS"] = _normalize_ccs_metadata_value(psm.metadata["CCS"])
 
     if needs_target and not all_has_targets:
         raise IM2DeepError("PSMList must contain 'ion_mobility' or 'CCS' metadata for all PSMs.")

im2deep/calibration.py

@@ -332,7 +332,7 @@ def get_charge(pf):
         # Extract CCS from metadata if it's not a direct column
         if "CCS" not in target_work.columns and "metadata" in target_work.columns:
             target_work["CCS"] = target_work["metadata"].apply(
-                lambda x: x.get("CCS", np.nan) if isinstance(x, dict) else np.nan
+                lambda x: float(x.get("CCS")) if isinstance(x, dict) and x.get("CCS") is not None else np.nan
             )
 
         source_work["peptide_key"] = source_work["peptidoform"].apply(get_peptide_key)
@@ -434,7 +434,7 @@ def get_charge(pf):
 
         if "CCS" not in target_work.columns and "metadata" in target_work.columns:
             target_work["CCS"] = target_work["metadata"].apply(
-                lambda x: x["CCS"] if "CCS" in x else np.nan
+                lambda x: float(x["CCS"]) if isinstance(x, dict) and "CCS" in x else np.nan
             )
 
         source_work["peptide_key"] = source_work["peptidoform"].apply(get_peptide_key)

im2deep/core.py

@@ -128,11 +128,7 @@ def predict_and_calibrate(
         )
 
     if not calibration.is_fitted:
-        if psm_df_reference is None:
-            raise ValueError(
-                "Reference PSM list must be provided for calibration fitting when using a custom "
-                "calibration object."
-            )
+        calibration_reference = psm_df_reference if psm_df_reference is not None else psm_df_cal
         LOGGER.info("Fitting calibration...")
         if any(psm_list_cal["is_decoy"]):
             LOGGER.warning(
@@ -141,7 +137,7 @@ def predict_and_calibrate(
             )
         calibration.fit(
             psm_df_cal,
-            psm_df_reference,
+            calibration_reference,
             multi=multi,
         )
     else:

im2deep/exceptions.py

@@ -8,6 +8,8 @@
 -------
 IM2DeepError
     Base exception class for all IM2Deep-related errors.
+PSMMetadataError
+    Exception raised when PSM metadata has an invalid type or value.
 CalibrationError
     Exception raised when calibration-related errors occur.
 """
@@ -36,6 +38,17 @@ class IM2DeepError(Exception):
     pass
 
 
+class PSMMetadataError(IM2DeepError, ValueError):
+    """
+    Exception raised when PSM metadata violates IM2Deep expectations.
+
+    This is used when metadata is structurally present but contains invalid
+    values for fields that IM2Deep needs to parse, such as non-numeric CCS.
+    """
+
+    pass
+
+
 class CalibrationError(IM2DeepError):
     """
     Exception raised when calibration-related errors occur.

tests/test_utils.py

@@ -6,7 +6,7 @@
 from psm_utils import PSMList
 
 from im2deep._io_helpers import parse_input, validate_psm_list
-from im2deep.exceptions import IM2DeepError
+from im2deep.exceptions import IM2DeepError, PSMMetadataError
 from im2deep.utils import ccs2im, im2ccs
 
 
@@ -26,14 +26,22 @@ def test_validate_psm_list_with_ccs(self, sample_psm_list_with_ccs):
         for psm in result:
             assert psm.metadata is not None
             assert "CCS" in psm.metadata
-            # CCS should always be stored as float
-            assert isinstance(psm.metadata["CCS"], float)
+            # PSM metadata is string-valued; CCS should be normalized to a numeric string.
+            assert isinstance(psm.metadata["CCS"], str)
+            float(psm.metadata["CCS"])
 
     def test_validate_psm_list_missing_ccs(self, sample_psm_list):
         """Test validation fails when CCS values are required but missing."""
         with pytest.raises(IM2DeepError, match="ion_mobility.*CCS.*metadata"):
             validate_psm_list(sample_psm_list, needs_target=True)
 
+    def test_validate_psm_list_invalid_ccs_value(self, sample_psm_list_with_ccs):
+        """Test validation fails with a custom metadata error for invalid CCS values."""
+        sample_psm_list_with_ccs[0].metadata["CCS"] = "not-a-number"  # type: ignore[index]
+
+        with pytest.raises(PSMMetadataError, match="Invalid CCS metadata value"):
+            validate_psm_list(sample_psm_list_with_ccs, needs_target=True)
+
     def test_validate_psm_list_empty(self):
         """Test validation with empty PSMList."""
         empty_list = PSMList(psm_list=[])