v0.7.1

Added

• --polygon-format=GeometryCollectionLegacy format similar to GeometryCollection, but with integer cell IDs. Used for compatibility with Xenium Ranger. This option will be deprecated after Xenium Ranger is updated.

Fixed

• CLI argument parsing

Changed

• polygons.json now has polygons for all cells in the dataset. Cells with 1 or 2 points now have polygons with 4 vertices.
• Polygon vertices are now slightly shifted farther from the cell center to avoid molecules on the border.

v0.7.0

Added

• New algorithm for NCV estimation based on Random Indexing. It is used by default now.
• Support of Parquet format for input molecules (can directly use transcripts.parquet from Xenium)
• Now, in case of continuous z-stack (like Xenium), 3D polygons are estimated by binning the z-stack into 20 slices.
• Added support for string IDs for prior segmentation. To specify the label for non-assigned molecules, use --config.segmentation.unassigned_prior_label.
• New plotting configuration options (see example_config.toml)

Removed

• no-ncv-estimation was removed, as the NCV algorithm is fast and memory-efficient now
• The Dirichlet sampling was removed, as it slowed-down the algorithm significantly, but didn't improve the segmentation quality
  • Correspondingly, parameters new_component_weight and new_component_fraction were removed

Fixed

• Improved multithreading
• Optimized the algorithm performance
• Improved algorithm for polygon estimation: fixed bugs and reduced overlaps
• Fixed a bug in the split step, improved cell continuity

Changed

• Cell IDs in polygon GeoJSON are now strings to match segmentation.csv
• The whole polygon.json format was changed from GeometryColection to FeatureCollection to match 10x format.
• Polygons are now saved by default using the 10x FeatureCollection format. Parameter save-polygons is replaced with polygon-format=FeatureCollection. Set it to GeometryCollection to save polygons in the format from Baysor v0.6 or to none to prevent saving polygons.
• 2D and 3D polygons are now stored in separate files (polygons_2d/3d.json)

Version 0.6.2

Fixed

• Fixed wrong parameter interpretation in neighborhood_count_matrix()
• Fixed incompatibility with julia-1.9 (#78)
• Fixed boundary_polygons_from_grid error (#82)
• Fixed a problem with huge integers in cell ids (#83)

Version 0.6.1

[0.6.1] — 2023-05-11

Fixed

• Fixed a bug with prior segmentation loading
• Fixed a bug with compartment genes

Version 0.6.0

[0.6.0] — 2023-04-20

Added

• New output cell QC parameters avg_assignment_confidence, max_cluster_frac and lifespan
• Segmented cells are now saved to loom instead of TSV. To return an old behavior, use count-matrix-format="tsv"
• Minimal multi-threading (see README)

Removed

• iters and n-cells-init parameters were removed from the CLI shortcuts. To change them, use the config or --config.segmentation.iters and --config.segmentation.n_cells_init parameters (see 'Advanced configuration section in the readme').

Changed

• Breaking changes in config file structure and CLI
• Greatly improved responsiveness of the CLI and simplified installation process
• Major refactoring of the code
• Various performance improvements
• Faster and more precise algorithm for estimating boundary polygons. Now each cell has exactly one polygon in the output GeoJSON.
  • For method details see Awrangjeb, 2015, it's pretty similar.
  • Closes #15 and #41, potentially also #32 and #37.
• Using sparse PCA for NCV estimation on large datasets
• baysor segfree output is now fully compatible with loom v3 format
• Cells and NCVs now have IDs in the format {type}{run_id}-{cell_id}, where type is C for cells and V for NCVs, and run_id is a unique ID of Baysor run
• --save-polygons now works regardless of -p

v0.5.2

[0.5.2] — 2022-06-29

Fixed

• Fixed some package versions, dependencies should cause fewer bugs now
• Fixed some bugs
• Fixed random seed for all CLI runs
• Adjusted some CLI parameters

Added

• Added segfree run option to extract NCV vectors

Version 0.5.1

Changed

• Slightly optimized compilation time
• Minor updates of core formulas

Added

• CLI parameter no-ncv-estimation to disable estimation of NCVs
• MacOS build

Version 0.5.0

Changed

• scale-std now can be specified from CLI parameters
• Several bugs fixed
• Allow missing genes in the input data
• Updates in the core algorithm
• All diagnostic plots were updated

Added

• exclude-genes option that removes genes from the data frame before segmentation.
• Segmentation of compartments based on the list of compartment-specific genes
• Using information about compartment per molecule in the segmentation algorithm when available
• 3D segmentation

Removed

• The data can not be split by frames anymore. This functionality didn't work well previously and was hard to maintan.

Version 0.4.3

Changed

• Fixed the CLI installation bug
• More information in logging
• Better initialization for molecule clustering. This should improve cell separability a lot!
• Some memory optimization

Added

• Estimating scale when prior segmentation is provided as a CSV column
• Added the option --save-polygons=GeoJSON to save cell boundary polygons in the GeoJSON format

Version 0.4.2

Changed

• Fixed Makefile julia version
• Improved polygon visualization
• Regressed to Plots 1.6.0 because of the performance issues
• Fixed docker build
• Small bug fixes
• min-transcripts-per-center is renamed to min-molecules-per-segment and is working now
• Fixed visualization of prior segmentation