agentleFS
NeuroAIHubNeuroAIHub/awesome_cognitive_and_neuroscience_skills/skills/netneurotools-guide/SKILL.md

netneurotools-guide

Skill39 starsChanged 9 months agoMITView on GitHub

Domain-validated guidance for network neuroscience analysis using netneurotools: datasets, brain network metrics, connectivity consensus, modularity, spatial statistics, null models, and cortical surface visualization. Use this skill whenever the user works with brain connectivity matrices, connectomes, graph theory on brain networks, parcellated brain data (Schaefer, Cammoun, Desikan-Killiany), cortical surface templates (fsaverage, fsLR, CIVET, Conte69), network communication metrics, null model generation, spatial autocorrelation on brain maps, community detection in connectomes, or surface-based visualization with PyVista/PySurfer. Also trigger when the user mentions netneurotools, netneurolab, structure-function coupling, network neuroscience, brain graph analysis, or needs to fetch neuroimaging atlases and templates.

  • Installs packages

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 0 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Weight-to-distance conversion: must be done before calling distance_wei_floyd or search_information.

    line 192
  2. Very specific:

    pv_plot_surface data format: When hemi='both', vertex_data can be a tuple (left, right) or a single concatenated array.

    line 200
  3. Very specific:

    Data directory: All fetch functions default to ~/nnt-data.

    line 202
  4. Very specific:

    match_length_degree_distribution recommended nswap: Use nswap = nnodes * 20 for adequate randomization.

    line 204
  5. Very specific:

    strength_preserving_rand_sa frac parameter: must be between 0 and 1. It controls temperature decrease per annealing stage.

    line 206

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

NetNeuroTools Guide

Purpose

This skill encodes the complete API and recommended workflows for netneurotools, a Python toolbox for network neuroscience developed by the Network Neuroscience Lab (netneurolab). It covers dataset fetching, brain connectivity metrics, network randomization and null models, community detection, spatial autocorrelation statistics, parcellation interface utilities, and cortical/subcortical surface visualization.

When to Use This Skill

Activate when the user:

  • Works with brain connectivity matrices (structural or functional)
  • Needs network communication metrics (shortest path, navigation, communicability, search information, diffusion efficiency)
  • Wants to generate null/surrogate networks preserving degree, strength, or distance distributions
  • Performs community detection or consensus clustering on brain networks
  • Computes spatial autocorrelation (Moran's I, Geary's C, Lee's L) on parcellated brain data
  • Fetches standard neuroimaging templates (fsaverage, fsLR, CIVET, Conte69) or atlases (Schaefer, Cammoun, Pauli)
  • Needs to convert between vertex-level and parcel-level brain data
  • Visualizes data on cortical surfaces using PyVista or PySurfer
  • Mentions netneurotools, netneurolab, or any function from this toolbox
  • Performs structure-function coupling, assortativity analysis, or dominance analysis
  • Simulates atrophy spreading on brain networks (SIR model)

Reference Files (Progressive Disclosure)

TopicFileWhen to Read
Datasetsreferences/datasets.mdUser fetches templates, atlases, or project datasets
Network Metricsreferences/metrics.mdUser computes communication, assortativity, spreading, or statistical network metrics
Networksreferences/networks.mdUser builds consensus connectivity, randomizes networks, or thresholds graphs
Statisticsreferences/stats.mdUser runs permutation tests, correlations, residualization, or dominance analysis
Spatial & Modularityreferences/spatial-modularity.mdUser computes spatial autocorrelation or performs community detection
Interface & Plottingreferences/interface-plotting.mdUser converts parcels/vertices, handles CIFTI/GIFTI files, or plots on cortical surfaces

Installation

pip install netneurotools

# For PyVista surface plotting (recommended)
pip install netneurotools[pyvista]

# For PySurfer surface plotting (legacy)
pip install netneurotools[pysurfer]

# For numba acceleration
pip install netneurotools[numba]

Core dependencies: numpy>=1.16, scipy>=1.4.0, scikit-learn, matplotlib, nibabel>=3.0.0, nilearn, bctpy, tqdm, neuromaps

Overview Pipeline

1. Fetch data       --> netneurotools.datasets (templates, atlases, connectomes)
2. Build networks   --> netneurotools.networks (consensus, thresholding)
3. Analyze metrics  --> netneurotools.metrics (communication, assortativity)
4. Null models      --> netneurotools.networks (randomization, surrogates)
5. Statistics       --> netneurotools.stats (permutation tests, dominance)
6. Spatial stats    --> netneurotools.spatial (Moran's I, Geary's C, Lee's L)
7. Modularity       --> netneurotools.modularity (consensus clustering)
8. Visualize        --> netneurotools.plotting (cortical surfaces, heatmaps)

Quick Start

Fetch Atlas and Template

from netneurotools.datasets import fetch_schaefer2018, fetch_fsaverage_curated

# Fetch Schaefer 400-parcel atlas in fsaverage space
parc = fetch_schaefer2018('fsaverage')['400Parcels7Networks']
# parc is a SURFACE namedtuple with fields .L and .R

# Fetch curated fsaverage surfaces
surfaces = fetch_fsaverage_curated('fsaverage5')
# surfaces has keys: 'white', 'pial', 'inflated', 'sphere', 'medial', 'sulc', 'vaavg'
# Each value is a SURFACE namedtuple with fields .L and .R

Consensus Functional Connectivity

from netneurotools.networks import func_consensus
import numpy as np

# data: (N_nodes, T_timepoints, S_subjects) array
consensus = func_consensus(data, n_boot=1000, ci=95, seed=42)

Community Detection

from netneurotools.modularity import consensus_modularity
import numpy as np

# adjacency: (N, N) non-negative connectivity matrix
consensus, Q_all, zrand_all = consensus_modularity(
    adjacency, gamma=1.5, repeats=100, seed=1234
)

Generate Distance-Preserving Surrogates

from netneurotools.networks import match_length_degree_distribution

newB, newW, nr = match_length_degree_distribution(
    W, D, nbins=10, nswap=1000, seed=42
)

Permutation Test for Correlation

from netneurotools.stats import permtest_pearsonr, make_correlated_xy

x, y = make_correlated_xy(corr=0.3, size=100, seed=42)
r, p = permtest_pearsonr(x, y, n_perm=5000, seed=42)

Spatial Autocorrelation

from netneurotools.spatial import morans_i

I = morans_i(annotation_vector, spatial_weight_matrix)

Plot on Cortical Surface (PyVista)

from netneurotools.plotting import pv_plot_surface
import numpy as np

data_L = np.random.random((10242,))
data_R = np.random.random((10242,))
pl = pv_plot_surface(
    (data_L, data_R),
    template="fsaverage5",
    surf="inflated",
    cmap="viridis",
    lighting_style="plastic",
    jupyter_backend="static",
)

Plot Parcellated Data (Shortcut)

from netneurotools.plotting import pv_plot_parcellated_data
import numpy as np

data = np.random.rand(400)
pl = pv_plot_parcellated_data(data, 'schaefer400x7', template='fsaverage')

Key Data Structures

StructureDescriptionFields
SURFACEnamedtuple for hemisphere file pairs.L, .R (left/right hemisphere paths)
sklearn.utils.BunchDict-like object returned by fetch functionsVaries per function
FREESURFER_IGNORELabels to ignore in FreeSurfer parcellations["unknown", "corpuscallosum", "Background+FreeSurfer_Defined_Medial_Wall"]
PARCIGNORELabels to ignore in parcellation operations["unknown", "corpuscallosum", "Background+FreeSurfer_Defined_Medial_Wall", "???", "Unknown", "Medial_wall", "Medial wall", "medial_wall"]

Core Modules Quick Reference

ModuleKey FunctionsPurpose
datasetsfetch_fsaverage, fetch_schaefer2018, fetch_cammoun2012, fetch_conte69, fetch_famous_gmatFetch templates, atlases, connectomes
metricsdistance_wei_floyd, navigation_wu, communicability_wei, search_information, mean_first_passage_time, assortativity_und, simulate_atrophyNetwork communication and properties
networksfunc_consensus, struct_consensus, match_length_degree_distribution, strength_preserving_rand_saBuild consensus, generate null models
statspermtest_pearsonr, efficient_pearsonr, residualize, get_dominance_statsStatistical testing and regression
spatialmorans_i, gearys_c, lees_l, local_morans_i, local_gearys_c, local_lees_lSpatial autocorrelation
modularityconsensus_modularity, find_consensus, zrand, get_modularityCommunity detection and evaluation
interfacevertices_to_parcels, parcels_to_vertices, load_surf_parc_file, deconstruct_ciftiFormat conversion
plottingpv_plot_surface, pv_plot_parcellated_data, pv_plot_subcortex, plot_mod_heatmapVisualization

Common Pitfalls

  1. SURFACE fields are .L and .R, not .lh and .rh. Use surface.L and surface.R to access hemisphere paths.
  2. Weight-to-distance conversion must be done before calling distance_wei_floyd or search_information. Common transform: D = -np.log(W / (np.max(W) + 1)).
  3. Minimum permutation p-value is 1 / (n_perm + 1). With n_perm=1000, the smallest p-value is ~0.001.
  4. consensus_modularity requires non-negative input. Louvain cannot handle negative weights. Set negatives to zero: A[A < 0] = 0.
  5. struct_consensus hemiid encoding: 0 = right hemisphere, 1 = left hemisphere.
  6. pv_plot_surface data format: When hemi='both', vertex_data can be a tuple (left, right) or a single concatenated array. Data length must match template vertex count.
  7. Data directory: All fetch functions default to ~/nnt-data. Override with data_dir= parameter or set NNT_DATA environment variable.
  8. match_length_degree_distribution recommended nswap: Use nswap = nnodes * 20 for adequate randomization.
  9. strength_preserving_rand_sa frac parameter must be between 0 and 1. It controls temperature decrease per annealing stage.
  10. parcels_to_vertices and vertices_to_parcels support .annot, .gii, and .dlabel.nii parcellation files. For .dlabel.nii, pass a single file path; for .annot or .gii, pass a tuple of (left, right) paths.
  11. Numba acceleration is available for spatial stats, weighted correlation, and some metrics. Install numba for significant speedups on large datasets.
  12. Headless rendering with PyVista: set os.environ["VTK_DEFAULT_OPENGL_WINDOW"] = "vtkOSOpenGLRenderWindow" before importing pyvista.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.