simnibs-analyze config

Validated against PipelineConfig before any work starts, so a malformed file fails immediately with a precise message.

config_analyze_example.yaml

Top-level keys

Key	Type	Default	Description
subjects	list of str	—	Subject IDs
stim_conditions	list of str	—	Conditions to process. Every entry needs a matching key in target_generation.rois
mode	list	—	Any of simulation, optimization
space	str	mni	mni or native. Drives the space-<...> output suffix
running	mapping	see below	Output-collision policy
paths	mapping	—	Input and output locations
target_generation	mapping	—	ROI definitions
preprocessing	mapping	see below	Smoothing and outlier removal
feature_extraction	mapping	see below	Which metrics to compute
analysis	mapping	see below	Group statistics and clustering

running

Key	Type	Default	Description
if_exists	str	skip	skip, overwrite, or error

paths

Key names are shared with the viz config, so a paths block can be copied between the two.

Key	Type	Description
sim_base	path	Simulation / optimization results — {sub}/simulations/
seg_base	path	Segmentation results — {sub}/m2m_{sub}/
out_root	path	Required. Where CSVs and statistics are written

Without seg_base you get intra-ROI stats only: no focality ratio, no clustering, and native-space runs cannot warp MNI targets.

target_generation

Key	Type	Default	Description
radius_mm	float	10.0	Sphere radius, must be > 0
rois	mapping	—	ROI name → definition

Coordinates are always MNI

coords are MNI millimetres regardless of space. In native, they are warped onto the subject grid using the field in seg_base/{sub}/m2m_{sub}/toMNI/. Atlas parcels are warped the same way.

ROI names cannot contain underscores

Use hyphens — ips-left, not ips_left — they collide with the space-<value> filename convention. SimNIBS folders on disk often use underscores anyway; both spellings are tried automatically.

Each ROI is one of two methods, discriminated on method:

rois:
  AFFT:
    method: sphere
    coords: [28, -8, 54]        # exactly 3 floats, MNI mm

  HA-fef:
    method: atlas
    atlas: harvard-oxford       # harvard-oxford | aal | destrieux
    regions: "Precentral Gyrus" # str or list of str

Key	Applies to	Description
coords	sphere	[x, y, z] in MNI mm
atlas	atlas	harvard-oxford, aal, or destrieux
regions	atlas	One label or a list of labels

preprocessing

Key	Type	Default	Description
smooth_fwhm	float	2.0	Smoothing FWHM in mm, >= 0
outlier_method	str	iqr	iqr or zscore
portion	float	null	Fraction of voxels to keep, in (0, 1]

feature_extraction

Key	Type	Default	Description
metrics	list of str	[mean, median, std, min, max]	Statistics computed per ROI

analysis

Key	Type	Default	Description
metric	str	mean	Column used for group statistics
subject_col	str	subject	Subject column in the features CSV
condition_col	str	condition	Condition column
clustering.method	str	mean	Reads efield_ratio_<method>
clustering.specificity_threshold	float	1.5	Ratio above which a simulation is specific, > 0
clustering.intensity_col	str	mean	Column whose group median splits high from low

Clusters are labelled specific_high, specific_low, diffuse_high, diffuse_low.

Minimal example

subjects: ["0001", "0002"]
stim_conditions: [AFFT]
mode: [simulation]
space: native

target_generation:
  radius_mm: 10.0
  rois:
    AFFT:
      method: sphere
      coords: [28, -8, 54]

paths:
  sim_base: /data/derivatives/simnibs-simu
  seg_base: /data/derivatives/simnibs-preps
  out_root: /data/derivatives/results

Checking a config

python -m simnibs_analyze._config_schema_analyze --config my-config.yaml