PreciseKids — Individualized Topography

This repository contains MATLAB, Python, R, and shell scripts supporting the following work:

Individualized network topography in pre-adolescent children and adults using naturalistic precision fMRI

First-author and script creation: Shefali Rai Under Review BioRxiv DOI: https://doi.org/10.64898/2026.03.05.709899

This repository covers individualized network topography analyses using naturalistic precision fMRI data, including generation of individual-level network maps, comparison with HCP templates, surface area gradiation, and network similarity between adults and children.

---

Repository Structure

PreciseKids_IndividualizedTopography/
├── scripts/        # MATLAB, Python, R, and shell analysis scripts
├── revisions/      # Revision analysis scripts (age, motion, and data length)
├── data/           # HCP overlap network files (as dscalars)
└── README.md

---

Requirements

• MATLAB R2021b or later
• Python 3 (for vertex-wise proportion assignment)
• R (for VertexwiseR TFCE analyses)
• Connectome Workbench (wb_view, wb_command) for visualization
• cifti-matlab and gifti-1.6 in your MATLAB path

---

Script Descriptions

Network Map Generation

MatchedCensoring_NetworkMaps_Allsubs.m Censors data across all tasks and sessions, then uses a sliding-window approach to randomly draw matched sections of usable (good) and censored data equally across sessions into two independent samples. Note: since sampling is still within-session, this is split-half.

createMaxDice_WinnerTakeAll.m Creates a group-average network map using a winner-take-all approach across participants. Used to create children and adult average maps.

HCP Template Comparisons

HCP_overlapbetweentemplates.m Quantifies the overlap between the Dworetsky HCP adult template and the HCP 8-9 year old template.

HCP_overlaptemplate_creation.m Creates the overlap map between the two HCP templates.

HCPOverlap_AllDiceValues_ExtractNetworkValues.m Extracts the full range of Dice overlap values per network for each participant.

Network Proportion & Surface Area

EachVertex_ProportionAssignment_HCPOverlap.py (Python) Computes child-minus-adult network assignment proportions at each vertex and outputs a differences text file used by NetworkPropDiff_Sample1.m.

NetworkPropDiff_Sample1.m For each vertex, computes the proportion of children and adults assigned to each network, then visualizes the absolute difference as a .dscalar map. Uses output from EachVertex_ProportionAssignment_HCPOverlap.py.

SurfaceArea_Gradiation.m Computes surface area and HCP overlap network assignment for each participant to generate gradiation (density) maps across all individuals.

---

Revision Analyses

Scripts in revisions/ are added for effects of head motion, split analyses in the main manuscript by motion groups, and impact of data quantity as varying amounts of data length/scan time. Motion groups are LMA (low-motion adults), LMC (low-motion children), and HMC (high-motion children).

Surface Area

SurfaceArea_AgeByNetwork.R Within-motion-group age-by-network surface-area models.

Density and Proportion Maps

CreateMotionGroup_DensityMaps.py (Python) Per-network group density maps for each motion group.

MakeProportion_FromDensity_60min.py (Python) Converts density maps to proportion (density/N) so motion groups of different sizes can be compared.

Figure3_MotionGroups_DensityMaps.m / Figure3_MotionGroups_ProportionMaps.m Apply Connectome Workbench palettes to the density and proportion maps for figures.

FamilyMatched_LMA10_Figure_and_Table.py (Python) Family-matched low-motion comparison (low-motion children vs their parents; n=10 to match children n) of network spatial spread.

Assignment Confidence (Entropy) and Reliability

Figure5_PanelA_GroupMeanEntropy.py (Python) Group-mean vertex-wise entropy maps for children and adults at 9 and 60 minutes.

LMMStats_Entropy_LMAvsLMC_60min_balanced.py (Python) Per-vertex mixed model of entropy between low-motion children and low-motion adults (n=10 each).

Figure5_Entropy_NMIcontrol.py (Python) Mean entropy controlled for topographic reliability (30-minute NMI) and head motion.

Entropy_IncreasingData_Stats.R Mean entropy across data lengths (9, 15, 30, 60 minutes) by motion group.

NMI_IncreasingData_Stats.R NMI reliability across data lengths by motion group.

Network Fragmentation

NetworkFragmentation_9vs60min.py (Python) Counts spatially contiguous clusters (fragments >= 15 mm2) per network at 9 and 60 minutes, for every participant.

Fragmentation_PairedTests_Figure.R Mixed model for fragment count (per net) between 9 and 60 minutes, with FDR corr.

Exemplar_NetworkConjunction_9vs60.py (Python) Conjunction maps for exemplar participants showing where a network is stable across both data lengths, present only at 9 minutes, or only at 60 minutes. Only visualized 1 exemplar child for the manuscript.

Similarity

FamilialSimilarity_SexDyad.R / WithinGroup_SexDyad.R Related vs unrelated (and within-group) Dice similarity, recoded by new sex dyad coding as per revisions (MM / FF / MF).

Split-Half Sampling (HPC)

MatchedCensoring_NetworkMaps_Revisions9min_SplitHalf.m / 15min / 30min Matched-censoring split-half network maps at 9, 15, and 30 minutes. arc_cluster_scripts/ contains SLURM and shell job scripts.

---

Notes on Visualization

• Open .dscalar.nii output files in Connectome Workbench (wb_view)
• Load .spec file from your data_path/subject/MNINonLinear/fsaverage_LR32k folder
• Set surface view to inflated.32k_fs_LR.surf.gii for both hemis

---

Citation

If you use these scripts, please cite our bio archive doi: https://doi.org/10.64898/2026.03.05.709899