Command-line reference

Start with Cell segmentation, Preview or Segmentation-free analysis for typical commands. These tables list the CLI options and defaults without a config file. Config values become defaults; explicit CLI flags override them.

baysor run --help
baysor preview --help
baysor segfree --help

Input and threads (all subcommands)

Each subcommand takes one required input: a CSV / Parquet molecule table or Xenium experiment.xenium manifest. See Input data.

Option	Default	Meaning
-c, --config	—	TOML file; use the space-separated form -c config.toml.
-m, --min-molecules-per-cell	—	Minimum molecules expected in a real cell. Must be positive (CLI or config); drives derived defaults.
-x, --x-column	x	x-coordinate column.
-y, --y-column	y	y-coordinate column.
-z, --z-column	z	Optional z-coordinate column.
-g, --gene-column	gene	Gene-name column.
--qv-column	qv	Quality-value column used by --min-qv.
--min-qv	-1	Drop molecules below this quality value; disabled unless set to a nonnegative value.
--force-2d	off	Ignore the z column.
--x-min, --x-max	−∞, +∞	Keep molecules within this x range.
--y-min, --y-max	−∞, +∞	Keep molecules within this y range.
--z-min, --z-max	−∞, +∞	Keep molecules within this z range.
-t, --threads	auto	Worker threads; OMP_NUM_THREADS, then physical CPU cores.

--help lists a subcommand's options. baysor --version prints the release version.

Cell segmentation (run)

The optional second input is a prior segmentation: a TIFF mask, a boundary table or :column_name from the molecule table.

Filtering

Option	Default	Meaning
--min-molecules-per-gene	1	Drop genes with fewer molecules.
--exclude-genes	empty	Comma-separated gene names or glob patterns (*, ?); quote patterns, e.g. --exclude-genes 'Blank*,MALAT1'.

These flags are run-only. The corresponding config keys also apply to preview and segfree.

Scale and convergence

Option	Default	Meaning
-s, --scale	—	Approximate cell radius. A positive scale disables scale estimation from the prior.
--scale-std	25%	Scale variation across cells: absolute value or percentage of scale. Replaced by the prior estimate when scale is estimated from a prior.
--iters	500	Maximum segmentation iterations.
--tol	0	Stop when fewer than this fraction of molecules change assignment over 20 consecutive iterations; 0 runs all iterations.
--n-cells-init	auto	Initial cell count; derived from molecule count and -m, reduced when a prior is available.

Clustering

Option	Default	Meaning
--cluster-method	mrf	mrf, louvain, leiden or none; ica_mrf is a legacy alias for mrf.
--n-clusters	auto	Exact count for mrf (default 4); merged target for louvain / leiden (default 10).
--cluster-resolution	1.0	Advanced overclustering resolution for graph methods.
--cluster-graph-k	15	NCV neighbors for graph clustering and NCV UMAPs.
--cluster-n-dims	20	NCV dimensions used by graph methods.
--cluster-basis-sample-size	100000	Maximum NCV basis anchors for graph methods.

Prior handling

Option	Default	Meaning
--prior-segmentation-confidence	0.2	Prior confidence in [0, 1]; see Prior behavior.
--unassigned-prior-label	0	Column-prior value meaning no assignment.

Output

Option	Default	Meaning
-o, --output	segmentation	Output directory.
--output-style	legacy	legacy or parquet; see Outputs.
--polygon-format	FeatureCollection	FeatureCollection, GeometryCollection, GeometryCollectionLegacy or none (case-insensitive). Legacy output only.
--count-matrix-format	loom	loom or tsv. Legacy output only.
-p, --plot	off	Write diagnostic_report.html and segmentation_plot.html.
--skip-ncv-color	off	Skip NCV color embedding; omits ncv_color from the molecule table.

Not implemented

--nuclei-genes and --cyto-genes accept comma-separated compartment-specific genes (default empty), but compartment segmentation is not implemented in C++. Setting either option makes run exit with an error.

Dataset preview (preview)

In addition to the shared options:

Option	Default	Meaning
-o, --output	preview.html	Output HTML file, not a directory.

Segmentation-free analysis (segfree)

In addition to the shared options:

Option	Default	Meaning
-k, --k-neighbors	auto	Neighbors per NCV; max(n_genes / 10, min_molecules_per_cell, 3) with integer division.
-o, --output	ncvs.loom	Output Loom file, not a directory.