Support homeXenium RangerRelease Notes
Release Notes for Xenium Ranger

Release Notes for Xenium Ranger

  • Required for reanalysis of Xenium Onboard Analysis v3.2 and upcoming v3.3 data (see table).
  • Updates file formats to match XOA v3.2:
    • Enables per-codeword localization plots for custom and boosted genes in Xenium Explorer v3.2.
    • Enables QC of boosted genes in analysis summary.
  • Multimodal cell segmentation algorithm: improved filtering of incorrect boundary-stain segmentations and merging segmentations of boundary-stain images.
  • Bug fixes:
    • Fixed missing aux_outputs/ directory and Image QC tab in the analysis summary.
    • Fixed misaligned Region Details plot in the analysis summary.
    • Fixed a resegment pipeline error that occurs when only --interior-stain or only --boundary-stain is disabled.
    • Fixed error when transcript-based --viz-polygons files are non-cyclic (missing closing vertex point) to enable importing results from Baysor v0.7.
  • Fixed a bug where pipeline errors due to the presence of unexpected files in the aux_outputs/ directory, such as AppleDouble format files created by MacOS.
  • Required for reanalysis of Xenium Onboard Analysis v3.0 data (see table).
  • Reanalyze pre-v3.0 data to convert to XOA v3.0 file formats that support scaled transcript view in Xenium Explorer v3.0. When reanalyzing XOA v3.0+ data, Xenium Ranger recomputes the transcript clusters in a different order so the scaled transcript view may look slightly different.
  • New xeniumranger rename subcommand for editing the region name and cassette name specified on instrument.
  • When only importing cell segmentations, Xenium Ranger no longer produces synthetic nucleus segmentations because visualization of a single polygon set is now supported in Xenium Explorer v3.0.
    • Importing transcript-based segmentation results with transcript assignments and boundaries will produce only one polygon set in the cells.zarr.zip.
    • Importing image-based segmentation results with cells will produce two polygon sets and masks in the cells.zarr.zip file, where the polygon set and mask that are usually reserved for nuclei will be empty.
    • Nuclear transcripts per 100 µm2 will be NaN because the nuclei boundaries are not defined. Use cellular transcripts per 100 µm2 instead to estimate transcript density.
  • Output format changes (will be added to upcoming XOA v3.1 release):
    • The cell_feature_matrix.zarr.zip is zstd compressed and includes a compressed sparse column (CSC) matrix representation for enhanced performance of certain features in future Xenium Explorer releases.
    • Upgraded experiment.xenium version to v5.1 with new fields: panel_designer_version, num_transcripts, num_transcripts_high_quality, and non_zero_matrix_entries
    • metrics_summary.csv values truncated to 15 decimal places.
  • General bug fixes and improvements:
    • Cell segmentation: Fixed a bug to guarantee every nucleus and cell segmentation mask is a single connected component. Cell segmentation algorithm matches algorithm in XOA v3.0.
    • (Note added 9/19/24) Cell segmentation: Improved multimodal cell segmentation by filtering incorrect segmentations of the boundary-stain images.
    • Robust to transcript-assignment data where one or more cells have no > Q20 transcripts. This fixes a bug in XR v2.0.1 that could result in erroneous transcript assignment to cells when the transcript-assignment CSV included cells with only low-quality transcripts (see this Knowledge Base article for help).
  • Fixed a bug that caused cell_ids to be misassigned in transcripts.csv.gz and transcripts.parquet when importing transcript assignments (e.g. from Baysor) with import-segmentation. This bug also resulted in erroneous cell-feature matrix files and erroneous counts in cells.csv.gz. Any datasets that were segmented with transcript-based methods, such as Baysor or Proseg, and used Xenium Ranger v2.0 to import transcript assignments, should re-run the analysis using Xenium Ranger v2.0.1.
  • Fixed a bug to ensure interior stain-based expansion extends all the way to the neighboring cell's boundary in resegment.
  • The expansion-distance parameter default is now 5 µm instead of 15 µm for the resegment and import-segmentation pipelines.
  • Segmentation algorithm updates for the resegment pipeline:
    • Reanalyze cell segmentation staining workflow data by adjusting the nuclear expansion distance and/or boundary or interior stain used for multimodal cell segmentation, the cell segmentation algorithm introduced in XOA v2.0. Suboptimal or unsuitable stains can be excluded from the segmentation algorithm by specifying the interior-stain or boundary-stain arguments.
    • Updated nucleus segmentation ML model to match update in XOA v2.0.
  • Updates for import-segmentation pipeline:
    • The xeniumranger resegment --resegment-nuclei option is deprecated and will be removed in future releases. resegment will always resegment nuclei from the input DAPI image. To use the nuclei segmentation from input data and adjust expansion distance, the preferred method is to run xeniumranger import-segmentation with --expansion-distance and --nuclei=cells.zarr.zip, which will use the nucleus segmentation masks from cells.zarr.zip file.
    • The --nuclei argument no longer needs to be used with --cells.
    • The --cells and --nuclei arguments support cells.zarr.zip format and will use the cell and nucleus masks for transcript assignment.
    • New metrics provided in metrics_summary.csv: "Number of cell non-simple polygons", "Number of cell multipolygons", "Number of nucleus non-simple polygons", "Number of nucleus multipolygons", "Segmented cell imported fraction", and "Segmented cell imported count". The "Number of cells from imported segmentation" is also reported in the analysis_summary.html Segmentation Methods table.
    • The "Percentage of cells without a nucleus" metric has been removed.
  • Interrupted runs can be resumed. Issuing the same xeniumranger command will now resume execution of the pipestance from the stage that originally failed. See troubleshooting page.
  • Fixed a bug where Xenium Ranger import-segmentation errored if GeoJSON inputs had the MultiPolygon type or Polygon type with holes. If Xenium Ranger encounters a polygon with a hole, it will remove the hole and retain only the exterior of the polygons (interior edges/segments are removed.) If type MultiPolygon is present, they will be ignored and will not be included in the analysis.
  • Fixed a bug to improve the method to import segmentation results generated from large image files (i.e., from IF or H&E images).
  • Fixed a bug to enable importing segmentation results for smaller regions of the original image data.
  • Supports reanalysis of Xenium Onboard Analysis data up to latest version (see version compatibility table).
  • The image-based import-segmentation --coordinate-transform option is now compatible with segmentation masks (previously limited to polygons).
  • Improved memory usage in the import-segmentation pipeline (see updated system requirements page).
  • Fixed bug where input masks with discontinuous cells or nuclei caused Xenium Ranger to occasionally crash. An analysis summary alert was introduced to flag rare scenarios where discontinuities may lead to splitting of imported cells or nuclei.
  • Several minor bug fixes in the analysis summary user interface.
  • Improved preflight checks.
  • Miscellaneous bug fixes.
  • Initial release.
  • Resegment Xenium data with customizable nucleus expansion distance and the latest 10x cell segmentation algorithm.
  • Import community-developed nucleus and cell segmentations in multiple popular formats (i.e., Baysor).
  • Selected the wrong gene panel when you started the Xenium Analyzer run? Simply fix the mistake by using relabel.
  • All pipelines produce the same output bundle as Xenium Onboard Analysis v1.6, so they are compatible with the same downstream visualization and analysis softwares (e.g., Xenium Explorer).
  • Backwards compatible with all previous Xenium Onboard Analysis versions.