Pipeline configuration

The atlas creation functions share common parameters that control pipeline behaviour. You can set these explicitly in each function call, globally via R options, or through environment variables.

Parameter hierarchy

Parameters resolve in this order:

Explicit argument — value passed directly to the function
R option — value from options()
Environment variable — value from Sys.getenv()
Default — built-in default value

This lets you set project-wide defaults while still overriding them for specific calls.

Available options

Parameter	R Option	Environment Variable	Default
`verbose`	`ggseg.extra.verbose`	`GGSEG_EXTRA_VERBOSE`	`TRUE`
`cleanup`	`ggseg.extra.cleanup`	`GGSEG_EXTRA_CLEANUP`	`TRUE`
`skip_existing`	`ggseg.extra.skip_existing`	`GGSEG_EXTRA_SKIP_EXISTING`	`TRUE`
`tolerance`	`ggseg.extra.tolerance`	`GGSEG_EXTRA_TOLERANCE`	`0.5`
`smoothness`	`ggseg.extra.smoothness`	`GGSEG_EXTRA_SMOOTHNESS`	`5`

Note: smoothness applies only to subcortical and tract pipelines. Cortical atlases use direct mesh projection and do not require smoothing.

Setting options in R

Use options() to set defaults for your R session:

options(
  ggseg.extra.tolerance = 0.5,
  ggseg.extra.cleanup = FALSE
)

atlas <- create_cortical_from_annotation(
  input_annot = c("lh.aparc.annot", "rh.aparc.annot"),
  output_dir = "my_atlas"
)

Verbosity

The verbose parameter controls progress messages during pipeline execution.

options(ggseg.extra.verbose = FALSE)

Sys.setenv(GGSEG_EXTRA_VERBOSE = "false")

Cleanup

The cleanup parameter controls whether intermediate files are removed after pipeline completion. Set it to FALSE to keep them for debugging:

options(ggseg.extra.cleanup = FALSE)

atlas <- create_subcortical_from_volume(
  input_volume = "aseg.mgz",
  output_dir = "my_atlas_files"
)

Skip existing

The skip_existing parameter lets you resume interrupted pipeline runs by reusing existing intermediate files:

options(ggseg.extra.skip_existing = FALSE)

options(ggseg.extra.skip_existing = TRUE)

Geometry parameters

The tolerance parameter controls the quality of 2D polygon geometry. Higher tolerance means fewer vertices — smaller file size, less detail.

For subcortical and tract pipelines, smoothness controls kernel smoothing of contour boundaries. Higher smoothness means rounder region boundaries.

options(ggseg.extra.tolerance = 0.5)

# smoothness only applies to subcortical/tract pipelines
options(ggseg.extra.smoothness = 15)

Environment variables

Environment variables are useful for CI pipelines, Docker containers, or settings that should persist across R sessions.

In .Renviron:

GGSEG_EXTRA_VERBOSE=false
GGSEG_EXTRA_CLEANUP=true
GGSEG_EXTRA_SKIP_EXISTING=true
GGSEG_EXTRA_TOLERANCE=0.5

In a shell:

export GGSEG_EXTRA_VERBOSE=false
export GGSEG_EXTRA_TOLERANCE=0.5
R -e "ggseg.extra::create_cortical_from_annotation(...)"

In Docker:

ENV GGSEG_EXTRA_VERBOSE=false
ENV GGSEG_EXTRA_CLEANUP=true

Overriding defaults

Explicit arguments always win:

options(ggseg.extra.cleanup = TRUE)

atlas <- create_cortical_from_annotation(
  input_annot = c("lh.aparc.annot", "rh.aparc.annot"),
  cleanup = FALSE
)

Recipes

Development and debugging

options(
  ggseg.extra.verbose = TRUE,
  ggseg.extra.cleanup = FALSE,
  ggseg.extra.skip_existing = FALSE
)

Production and CI

options(
  ggseg.extra.verbose = FALSE,
  ggseg.extra.cleanup = TRUE,
  ggseg.extra.skip_existing = TRUE
)

Iterating on tolerance

For cortical atlases, adjusting tolerance is the main tuning knob. Use 0 for maximum mesh fidelity, or higher values for smaller file sizes:

annot_files <- c("lh.myatlas.annot", "rh.myatlas.annot")

# High fidelity (no simplification)
atlas <- create_cortical_from_annotation(
  input_annot = annot_files,
  output_dir = "atlas_workdir",
  tolerance = 0
)

# Compact (more simplification)
atlas <- create_cortical_from_annotation(
  input_annot = annot_files,
  output_dir = "atlas_workdir",
  tolerance = 1
)