thyra 4.3.4__tar.gz → 4.4.0__tar.gz

{thyra-4.3.4 → thyra-4.4.0}/PKG-INFO

@@ -1,10 +1,11 @@
 Metadata-Version: 2.5
 Name: thyra
-Version: 4.3.4
+Version: 4.4.0
 Summary: A modern Python library for converting Mass Spectrometry Imaging (MSI) data into SpatialData/Zarr format - your portal to spatial omics
 Project-URL: Homepage, https://github.com/M4i-Imaging-Mass-Spectrometry/thyra
 Project-URL: Repository, https://github.com/M4i-Imaging-Mass-Spectrometry/thyra
 Project-URL: Documentation, https://M4i-Imaging-Mass-Spectrometry.github.io/thyra/
+Project-URL: Paper, https://doi.org/10.1021/jasms.6c00169
 Project-URL: Bug Tracker, https://github.com/M4i-Imaging-Mass-Spectrometry/thyra/issues
 Project-URL: Discussions, https://github.com/M4i-Imaging-Mass-Spectrometry/thyra/discussions
 Project-URL: Changelog, https://github.com/M4i-Imaging-Mass-Spectrometry/thyra/blob/main/CHANGELOG.md
@@ -172,6 +173,26 @@ MIT -- see [LICENSE](LICENSE).
 
 ## Citation
 
+If you use Thyra, please cite the paper:
+
+> Visvikis, T.; Vierdag, W.-M.; Marconato, L.; Birmpili, A.; Heeren, R. M. A.;
+> Cuypers, E. Native, Memory-Bounded Conversion of Mass Spectrometry Imaging
+> Data into the SpatialData Ecosystem. *J. Am. Soc. Mass Spectrom.* **2026**.
+> [doi:10.1021/jasms.6c00169](https://doi.org/10.1021/jasms.6c00169)
+
+```bibtex
+@article{visvikis2026thyra,
+  title   = {Native, Memory-Bounded Conversion of Mass Spectrometry Imaging Data into the {SpatialData} Ecosystem},
+  author  = {Visvikis, Theodoros and Vierdag, Wouter-Michiel and Marconato, Luca and Birmpili, Angeliki and Heeren, Ron M. A. and Cuypers, Eva},
+  journal = {Journal of the American Society for Mass Spectrometry},
+  year    = {2026},
+  doi     = {10.1021/jasms.6c00169}
+}
+```
+
+The same entry is in [CITATION.cff](CITATION.cff), which GitHub's "Cite this
+repository" button reads. To cite the software itself:
+
 ```bibtex
 @software{thyra2024,
   title = {Thyra: Modern Mass Spectrometry Imaging Data Conversion},

{thyra-4.3.4 → thyra-4.4.0}/README.md

@@ -119,6 +119,26 @@ MIT -- see [LICENSE](LICENSE).
 
 ## Citation
 
+If you use Thyra, please cite the paper:
+
+> Visvikis, T.; Vierdag, W.-M.; Marconato, L.; Birmpili, A.; Heeren, R. M. A.;
+> Cuypers, E. Native, Memory-Bounded Conversion of Mass Spectrometry Imaging
+> Data into the SpatialData Ecosystem. *J. Am. Soc. Mass Spectrom.* **2026**.
+> [doi:10.1021/jasms.6c00169](https://doi.org/10.1021/jasms.6c00169)
+
+```bibtex
+@article{visvikis2026thyra,
+  title   = {Native, Memory-Bounded Conversion of Mass Spectrometry Imaging Data into the {SpatialData} Ecosystem},
+  author  = {Visvikis, Theodoros and Vierdag, Wouter-Michiel and Marconato, Luca and Birmpili, Angeliki and Heeren, Ron M. A. and Cuypers, Eva},
+  journal = {Journal of the American Society for Mass Spectrometry},
+  year    = {2026},
+  doi     = {10.1021/jasms.6c00169}
+}
+```
+
+The same entry is in [CITATION.cff](CITATION.cff), which GitHub's "Cite this
+repository" button reads. To cite the software itself:
+
 ```bibtex
 @software{thyra2024,
   title = {Thyra: Modern Mass Spectrometry Imaging Data Conversion},

{thyra-4.3.4 → thyra-4.4.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "thyra"
-version = "4.3.4"
+version = "4.4.0"
 description = "A modern Python library for converting Mass Spectrometry Imaging (MSI) data into SpatialData/Zarr format - your portal to spatial omics"
 authors = [{ name = "Theodoros Visvikis", email = "t.visvikis@maastrichtuniversity.nl" }]
 maintainers = [{ name = "Theodoros Visvikis", email = "t.visvikis@maastrichtuniversity.nl" }]
@@ -243,6 +243,7 @@ dependencies = [
 Homepage = "https://github.com/M4i-Imaging-Mass-Spectrometry/thyra"
 Repository = "https://github.com/M4i-Imaging-Mass-Spectrometry/thyra"
 Documentation = "https://M4i-Imaging-Mass-Spectrometry.github.io/thyra/"
+Paper = "https://doi.org/10.1021/jasms.6c00169"
 "Bug Tracker" = "https://github.com/M4i-Imaging-Mass-Spectrometry/thyra/issues"
 Discussions = "https://github.com/M4i-Imaging-Mass-Spectrometry/thyra/discussions"
 Changelog = "https://github.com/M4i-Imaging-Mass-Spectrometry/thyra/blob/main/CHANGELOG.md"

{thyra-4.3.4 → thyra-4.4.0}/thyra/__init__.py

@@ -67,7 +67,7 @@ warnings.filterwarnings(
     category=FutureWarning,
 )
 
-__version__ = "4.3.4"
+__version__ = "4.4.0"
 
 #: Public name -> the module that defines it, imported on first access.
 _LAZY_ATTRS = {

{thyra-4.3.4 → thyra-4.4.0}/thyra/converters/spatialdata/base_spatialdata_converter.py

@@ -892,7 +892,7 @@ class BaseSpatialDataConverter(BaseMSIConverter, ABC):
         Otherwise, shapes use physical (micrometer) coordinates.
 
         **Footprints are two-dimensional, including on a multi-slice
-        volume.** A slice's depth is carried by the TIC image's ``Scale``
+        volume.** A slice's depth is carried by the TIC image's transform
         and by ``obs["spatial_z"]``; it is deliberately not also put on
         the polygon geometry.
 
@@ -1042,7 +1042,7 @@ class BaseSpatialDataConverter(BaseMSIConverter, ABC):
             half_y_um = self.pixel_size_y_um / 2
 
             # Footprints are flat, on every route including volumes. A
-            # slice's depth lives on the TIC image's Scale and in
+            # slice's depth lives on the TIC image's transform and in
             # obs["spatial_z"]; see the docstring for why it is not also
             # put on the geometry. Built through shapely's vectorised box
             # constructor -- one C call for the whole table instead of one

{thyra-4.3.4 → thyra-4.4.0}/thyra/converters/spatialdata/optical_image.py

@@ -1183,6 +1183,18 @@ class OpticalImages:
         transformation that maps MSI raster coordinates to optical image
         pixel coordinates.
         """
+        # A source that states its own registration (an mzPeak archive with
+        # a teach_points affine) hands it over whole.
+        stated = getattr(self.reader, "get_image_alignment", None)
+        alignment = stated() if callable(stated) else None
+        if alignment is not None:
+            self._alignment = alignment
+            logger.info(
+                "Optical alignment stated by the source, with "
+                f"{len(alignment.region_mappings)} region mappings"
+            )
+            return
+
         # Check if reader has FlexImaging-specific metadata
         if not hasattr(self.reader, "mis_metadata"):
             logger.debug("Reader does not have mis_metadata, skipping alignment")
@@ -1399,7 +1411,7 @@ class OpticalImages:
         primary = resolve() if callable(resolve) else None
         if primary is not None:
             self.primary_filename = Path(primary).stem.lower()
-            logger.info(f"Primary alignment image from .mis: {primary.name}")
+            logger.info(f"Primary alignment image named by the reader: {primary.name}")
 
     def _is_primary_optical(self, image_path: Path) -> bool:
         """Check if an image file is the primary alignment image from .mis.

{thyra-4.3.4 → thyra-4.4.0}/thyra/converters/spatialdata/root_attrs.py

@@ -232,12 +232,13 @@ class RootAttrsBuilder:
         Two variants are emitted depending on whether FlexImaging optical
         alignment was applied during conversion:
 
-        - No alignment (``global = micrometer``): the TIC image carries a
-          ``Scale(pixel_size_um)`` and pixel-polygon shapes are stored in
-          micrometers with ``Identity``. Both elements agree at
-          ``global``. ``pixel_size_um_x/y`` are filled with the MSI grid
-          pixel size, since "global" is in physical micrometers and there
-          is no canonical raster image other than the MSI itself.
+        - No alignment (``global = micrometer``): the TIC image carries
+          a half-cell ``Translation`` then ``Scale(pixel_size_um)``, and
+          pixel-polygon shapes are stored in micrometers with
+          ``Identity``. Both elements agree at ``global``.
+          ``pixel_size_um_x/y`` are filled with the MSI grid pixel size,
+          since "global" is in physical micrometers and there is no
+          canonical raster image other than the MSI itself.
 
         - With alignment (``global = pixel``): the TIC image carries an
           ``Affine`` mapping raster indices to optical-image pixels and
@@ -311,9 +312,10 @@ class RootAttrsBuilder:
         # the explicit 3x3 row-major affine from TIC raster indices to
         # "global" (the same mapping the TIC element's transform
         # expresses), so a consumer that reads only attrs still gets
-        # the full placement. In the pixel variant it takes TIC
-        # coordinates, where cell i spans [i, i + 1) and its centre
-        # i + 0.5 is the spot (D31).  `coordinate_offsets_px` preserves the
+        # the full placement. In both variants it takes TIC coordinates,
+        # where cell i spans [i, i + 1): its centre i + 0.5 lands on the
+        # spot (D31), or on obs["spatial_x"] = i * px in micrometres
+        # (D32).  `coordinate_offsets_px` preserves the
         # source's raw acquisition-index offsets, which 0-based
         # normalisation otherwise erases; `stage_offset_um` is their
         # physical equivalent, written only when "global" is in
@@ -325,8 +327,8 @@ class RootAttrsBuilder:
             ]
         else:
             global_cs["raster_to_global_affine"] = [
-                [px, 0.0, 0.0],
-                [0.0, py, 0.0],
+                [px, 0.0, -0.5 * px],
+                [0.0, py, -0.5 * py],
                 [0.0, 0.0, 1.0],
             ]
         offsets = self._source_coordinate_offsets()

{thyra-4.3.4 → thyra-4.4.0}/thyra/converters/spatialdata/streaming_converter.py

@@ -47,6 +47,8 @@ from anndata import AnnData
 from numpy.typing import NDArray
 from spatialdata.models import Image2DModel, Image3DModel, TableModel
 from spatialdata.transformations import Affine, Scale
+from spatialdata.transformations import Sequence as SequenceTransform
+from spatialdata.transformations import Translation
 from tqdm import tqdm
 
 from ...core.conversion_state import ConversionState
@@ -62,6 +64,17 @@ logger = logging.getLogger(__name__)
 _NOT_ACQUIRED = np.iinfo(np.int64).max
 
 
+def _cell_centred_scale(pitches: List[float], axes: Tuple[str, ...]) -> Any:
+    """Raster cells to micrometres, each cell centred on its own position.
+
+    Cell ``i`` spans ``[i, i + 1)``; moving it back half a cell first puts
+    its centre on ``i * pitch``, the position ``obs`` records (D32).
+    """
+    return SequenceTransform(
+        [Translation([-0.5] * len(axes), axes=axes), Scale(pitches, axes=axes)]
+    )
+
+
 class _TableUnit:
     """One table of the store in the making: a z plane, or the whole volume.
 
@@ -1127,6 +1140,11 @@ class StreamingSpatialDataConverter(BaseSpatialDataConverter):
         frame -- optical-image pixels when FlexImaging alignment is
         applied, physical micrometres otherwise, so that ``"global"``
         agrees with the pixel-polygon shapes.
+
+        SpatialData draws cell ``i`` over ``[i, i + 1)``. In micrometres
+        the cell is moved back half a step before it is scaled, so its
+        centre lands on ``obs["spatial_x"] = i * pixel_size_um``, where
+        the pixel's polygon is centred (D32).
         """
         n_z = self._dimensions[2]
         if unit.plane is None and n_z > 1:
@@ -1134,7 +1152,7 @@ class StreamingSpatialDataConverter(BaseSpatialDataConverter):
             # spacing (see BaseMSIConverter._resolve_z_spacing). Scale
             # pairs values with axis *names*, so ("x", "y", "z") against
             # a (c, z, y, x) image is deliberate.
-            transform: Any = Scale(
+            transform: Any = _cell_centred_scale(
                 [self.pixel_size_um, self.pixel_size_y_um, self.z_spacing_um],
                 axes=("x", "y", "z"),
             )
@@ -1155,7 +1173,7 @@ class StreamingSpatialDataConverter(BaseSpatialDataConverter):
                 output_axes=("x", "y"),
             )
         else:
-            transform = Scale(
+            transform = _cell_centred_scale(
                 [self.pixel_size_um, self.pixel_size_y_um], axes=("x", "y")
             )
         return Image2DModel.parse(

{thyra-4.3.4 → thyra-4.4.0}/thyra/converters/spatialdata/uns_assembler.py

@@ -37,6 +37,7 @@ every CSC scratch memmap the finalize path's frames still hold (issue
 
 import json
 import logging
+import numbers
 from dataclasses import dataclass
 from typing import (
     TYPE_CHECKING,
@@ -548,6 +549,11 @@ class UnsAssembler:
             and getattr(self.reader, "file_type", None) == "tdf"
         ):
             conversion_parameters["tdf_spectrum"] = str(tdf_spectrum)
+        # Peaks below it are not in the store, so a store made with one is
+        # not the source's whole signal, and must say so (D34).
+        intensity_threshold = getattr(self.reader, "intensity_threshold", None)
+        if isinstance(intensity_threshold, numbers.Real):
+            conversion_parameters["intensity_threshold"] = float(intensity_threshold)
         steps = [
             ProcessingStep(
                 name="conversion",

{thyra-4.3.4 → thyra-4.4.0}/thyra/core/base_reader.py

@@ -69,6 +69,7 @@ from numpy.typing import NDArray
 from ..metadata.types import ComprehensiveMetadata, EssentialMetadata
 
 if TYPE_CHECKING:
+    from ..alignment import AreaAlignmentResult
     from .base_extractor import MetadataExtractor
     from .frames import FrameScans
     from .mobility import MobilityAxis
@@ -78,6 +79,53 @@ from .mass_axis import validate_max_mass_axis_length
 
 logger = logging.getLogger(__name__)
 
+#: Two m/z of one spectrum this close, relative, are one peak listed twice.
+#: A source that lists one point per mobility scan repeats a peak's m/z
+#: (an archive's chunk bound can differ from the decoded value in the last
+#: bit), while the closest distinct peaks of any analyser are some 1e-6
+#: apart.
+SAME_PEAK_RELATIVE = 1e-12
+
+
+def summed_peak_mask(
+    mzs: NDArray[np.float64], intensities: NDArray[np.float64], threshold: float
+) -> NDArray[np.bool_]:
+    """Which points belong to a peak whose summed intensity reaches ``threshold``.
+
+    The intensity threshold's one unit (D34): a peak's intensity summed over
+    every point that lists its m/z. A spectrum that lists each m/z once is
+    tested point by point, as before; one that lists a peak once per mobility
+    scan keeps or drops all of that peak's points together, so the threshold
+    acts on the peak the store holds, not on one scan's share of it.
+
+    Args:
+        mzs: m/z of each point, in any order.
+        intensities: Intensity of each point.
+        threshold: The lowest summed intensity kept.
+
+    Returns:
+        A mask over the points, in their order.
+    """
+    if mzs.size < 2:
+        return intensities >= threshold
+    steps = np.diff(mzs)
+    if np.all(steps > SAME_PEAK_RELATIVE * np.abs(mzs[1:])):
+        # Strictly ascending with no repeat: every point is its own peak.
+        return intensities >= threshold
+    order = np.argsort(mzs, kind="stable")
+    ordered = mzs[order]
+    starts = np.concatenate(
+        ([True], np.diff(ordered) > SAME_PEAK_RELATIVE * np.abs(ordered[1:]))
+    )
+    # argsort puts NaN last; each NaN point is its own peak, never part of
+    # the largest real one.
+    starts |= np.isnan(ordered)
+    group = np.cumsum(starts) - 1
+    sums = np.bincount(group, weights=intensities[order])
+    keep = np.empty(mzs.size, dtype=bool)
+    keep[order] = sums[group] >= threshold
+    return keep
+
 
 class OpticalImageLabel(NamedTuple):
     """What a store calls an optical image whose file name is not the source's.
@@ -194,6 +242,20 @@ class BaseMSIReader(ABC):
         """
         return None
 
+    def get_image_alignment(self) -> Optional["AreaAlignmentResult"]:
+        """The registration the source states for its alignment image.
+
+        Default implementation returns None. A format whose file states
+        where each pixel sits on its alignment image overrides it; the
+        converter then places the pixels by it, as it places a FlexImaging
+        run's by its ``.mis``. The image is the one
+        :meth:`get_primary_optical_image_path` names.
+
+        Returns:
+            The alignment, or None when the source states none.
+        """
+        return None
+
     def get_optical_image_label(self, path: Path) -> Optional[OpticalImageLabel]:
         """Name an optical image this reader copied out of its source.
 
@@ -265,20 +327,29 @@ class BaseMSIReader(ABC):
     ) -> Tuple[NDArray[np.float64], NDArray[np.float64]]:
         """Apply intensity threshold filtering to spectrum data.
 
+        A peak is kept when its intensity, summed over every point that
+        lists its m/z, reaches the threshold (:func:`summed_peak_mask`).
+
         Args:
             mzs: m/z values array
             intensities: Intensity values array
 
         Returns:
-            Tuple of (filtered_mzs, filtered_intensities) with values below
-            threshold removed. Returns original arrays if no threshold is set.
+            Tuple of (filtered_mzs, filtered_intensities) with the peaks
+            below the threshold removed. Returns original arrays if no
+            threshold is set.
         """
         if self._intensity_threshold is None:
             return mzs, intensities
 
-        mask = intensities >= self._intensity_threshold
+        mask = summed_peak_mask(mzs, intensities, self._intensity_threshold)
         return mzs[mask], intensities[mask]
 
+    @property
+    def intensity_threshold(self) -> Optional[float]:
+        """The ``--intensity-threshold`` this reader applies, ``None`` for none."""
+        return self._intensity_threshold
+
     @staticmethod
     def map_mz_to_common_axis(
         mzs: NDArray[np.float64],

{thyra-4.3.4 → thyra-4.4.0}/thyra/metadata/constants.py

@@ -15,7 +15,7 @@ module re-exports every name here, so both spellings keep working;
 ``Thresholds``, the one genuine resampling number in it, stayed there.
 """
 
-from typing import Dict, Optional
+from typing import Dict, Iterable, Optional
 
 from ..errors import ConversionRefused
 
@@ -27,6 +27,10 @@ class ImzMLAccessions:
     CENTROID_SPECTRUM = "MS:1000127"
     PROFILE_SPECTRUM = "MS:1000128"
 
+    # Scan polarity (MS ontology)
+    POSITIVE_SCAN = "MS:1000130"
+    NEGATIVE_SCAN = "MS:1000129"
+
     # Binary data type (imzML ontology)
     CONTINUOUS_BINARY = "IMS:1000030"
     PROCESSED_BINARY = "IMS:1000031"
@@ -99,6 +103,36 @@ def normalize_spectrum_type(value: Optional[object]) -> Optional[str]:
     )
 
 
+#: The polarity each scan polarity term states.
+POLARITY_OF_ACCESSION: Dict[str, str] = {
+    ImzMLAccessions.POSITIVE_SCAN: "positive",
+    ImzMLAccessions.NEGATIVE_SCAN: "negative",
+}
+
+
+#: What an extractor records for a source that states both polarities:
+#: pyimzml's word for it. No polarity term matches it, so the document's
+#: field stays unset, and the file-level terms cannot fill it instead.
+MIXED_POLARITY = "mixed"
+
+
+def agreed_polarity(stated: Iterable[str]) -> Optional[str]:
+    """The polarity every statement of a source agrees on (D35).
+
+    ``stated`` holds ``"positive"`` or ``"negative"`` once for each place
+    the source states one: its file-level terms, a parameter group, a
+    spectrum. One value is the answer, and None means the source says
+    nothing. Both give :data:`MIXED_POLARITY`: the source alternated or
+    contradicts itself, which has no single truthful value.
+    """
+    values = set(stated)
+    if not values:
+        return None
+    if len(values) > 1:
+        return MIXED_POLARITY
+    return next(iter(values))
+
+
 class BinaryDataType:
     """Binary data type string constants."""
 

{thyra-4.3.4 → thyra-4.4.0}/thyra/metadata/extractors/imzml_extractor.py

@@ -2,7 +2,7 @@
 import gc
 import logging
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Tuple
+from typing import Any, Dict, List, Optional, Set, Tuple
 
 import numpy as np
 from numpy.typing import NDArray
@@ -13,9 +13,12 @@ from ...errors import ConversionRefused
 from ...utils.imzml_coordinate_base import coordinate_bases, recorded_offsets
 from ...utils.pyimzml_direct import read_spectrum_mzs_only
 from ..constants import (
+    MIXED_POLARITY,
+    POLARITY_OF_ACCESSION,
     BinaryDataType,
     ImzMLAccessions,
     SpectrumType,
+    agreed_polarity,
     normalize_spectrum_type,
 )
 from ..ontology.cache import ONTOLOGY
@@ -911,7 +914,7 @@ class ImzMLMetadataExtractor(MetadataExtractor):
 
     def _extract_acquisition_params(self) -> Dict[str, Any]:
         """Extract acquisition parameters from XML metadata."""
-        params = {}
+        params: Dict[str, Any] = {}
 
         # Extract pixel size with full XML parsing if not found in fast
         # extraction
@@ -921,6 +924,12 @@ class ImzMLMetadataExtractor(MetadataExtractor):
                 params["pixel_size_x_um"] = pixel_size[0]
                 params["pixel_size_y_um"] = pixel_size[1]
 
+        # Only when the file states one. "mixed" when it states both, so
+        # the builder cannot fall back to the file-level terms alone (D35).
+        polarity = self._extract_polarity()
+        if polarity is not None:
+            params["polarity"] = polarity
+
         # Add other acquisition parameters from imzmldict
         if hasattr(self.parser, "imzmldict") and self.parser.imzmldict:
             acquisition_keys = [
@@ -937,6 +946,59 @@ class ImzMLMetadataExtractor(MetadataExtractor):
 
         return params
 
+    def _extract_polarity(self) -> Optional[str]:
+        """The polarity every statement in the file agrees on (D35).
+
+        An imzML can state it in three places: the ``fileContent`` terms,
+        a referenceable parameter group, and each spectrum. Writers differ
+        (pyimzml's writer uses the group), so all three are read. The
+        spectra's terms come from the parser's own pass over them. A file
+        that states both values, anywhere, is left unset and says so.
+        """
+        stated: List[str] = []
+        metadata = getattr(self.parser, "metadata", None)
+        groups = [getattr(metadata, "file_description", None)]
+        param_groups = getattr(metadata, "referenceable_param_groups", None)
+        if isinstance(param_groups, dict):
+            groups.extend(param_groups.values())
+        for group in groups:
+            by_accession = getattr(group, "param_by_accession", None)
+            if isinstance(by_accession, dict):
+                stated.extend(
+                    polarity
+                    for accession, polarity in POLARITY_OF_ACCESSION.items()
+                    if accession in by_accession
+                )
+        stated.extend(
+            POLARITY_OF_ACCESSION[accession]
+            for accession in self._spectrum_polarity_accessions()
+        )
+        polarity = agreed_polarity(stated)
+        if polarity == MIXED_POLARITY:
+            logger.warning(
+                "%s states both positive and negative scan polarity; no "
+                "polarity is recorded.",
+                self.imzml_path.name,
+            )
+        return polarity
+
+    def _spectrum_polarity_accessions(self) -> Set[str]:
+        """The scan polarity terms any spectrum states itself.
+
+        Collected by the parser during its one pass over the spectra, when
+        it was built with ``include_spectra_metadata`` as the reader builds
+        it. A parser built without it gives none, and the polarity then
+        rests on ``fileContent`` and the groups.
+        """
+        fields = getattr(self.parser, "spectrum_metadata_fields", None)
+        if not isinstance(fields, dict):
+            return set()
+        return {
+            accession
+            for accession in POLARITY_OF_ACCESSION
+            if any(value is not None for value in fields.get(accession, ()))
+        }
+
     def _extract_instrument_info(self) -> Dict[str, Any]:
         """Extract instrument information from the instrumentConfiguration blocks.
 

{thyra-4.3.4 → thyra-4.4.0}/thyra/metadata/extractors/imzml_header_extractor.py

@@ -38,7 +38,7 @@ from __future__ import annotations
 
 import logging
 from pathlib import Path
-from typing import Any, Optional, Tuple, cast
+from typing import Any, Optional, Set, Tuple, cast
 
 from ...readers.imzml.header import ImzMLHeaderParser, scan_observed_mz_range
 from ..types import EssentialMetadata
@@ -148,6 +148,15 @@ class ImzMLHeaderExtractor(ImzMLMetadataExtractor):
         """The count the file declares, not one taken over coordinates."""
         return int(self.header.n_spectra or 0)
 
+    def _spectrum_polarity_accessions(self) -> Set[str]:
+        """The polarity terms the first spectrum states, as pyimzml reads it.
+
+        The head ends at the second spectrum. A conversion reads every
+        spectrum's terms, so the two differ only for a file whose later
+        spectra state the other polarity (D35).
+        """
+        return set(cast(ImzMLHeaderParser, self.parser).first_spectrum_polarity)
+
 
 def head_shortfall(parser: ImzMLHeaderParser) -> Optional[str]:
     """Why the head cannot describe this file, or ``None`` if it can.