Transcriptomic gradients (Allen Human Brain Atlas)

Overview

The first three principal gradients of transcriptomic variation across the cortex and subcortex, computed in Euclidean space from the six post-mortem brains of the Allen Human Brain Atlas (AHBA). These three spatial axes capture the directions along which gene expression varies most strongly — interpreted by Vogel et al. (2024) as candidate axes of morphogen diffusion during development, and offering an objective whole-brain orientation that is independent of cytoarchitecture or function. Distributed here in two MNI templates plus a CIFTI grayordinate version produced via registration fusion.

See README.md for the authoritative methods write-up, including the source-template ambiguity in the AHBA coordinates, the 6 mm smoothing kernel applied to absorb that uncertainty, and the CIFTI registration-fusion construction.

Primary reference. Vogel, J. W., Alexander-Bloch, A. F., Wagstyl, K., Bertolero, M. A., Markello, R. D., Pines, A., Sydnor, V. J., Diaz-Papkovich, A., Hansen, J. Y., Evans, A. C., Bernhardt, B., Misic, B., Satterthwaite, T. D., & Seidlitz, J. (2024). Deciphering the functional specialization of whole-brain spatiomolecular gradients in the adult brain. PNAS, 121(23), e2219137121. doi:10.1073/pnas.2219137121

The underlying transcriptomic atlas is from Hawrylycz et al. 2012 Nature (see Citations). Construction code is in create_gradients.py; the upstream pipeline lives at PennLINC/Vogel_PLS_Tx-Space.

Key images

Cortical surface (MNI152NLin2009cSym)	Axial montage (MNI152NLin2009cSym)

The three principal transcriptomic gradients in fmriprep space. The matching MNI152NLin6Asym (FSL-space) and isosurface renderings are also in png_images/; produced by visualize_contents.m. The CIFTI version is best viewed in Connectome Workbench.

How to load

Registered in load_image_set.m under the keyword 'transcriptomic_gradients':

[grad_obj, ~, ~] = load_image_set('transcriptomic_gradients');

Or load a specific template directly:

% FSL / MNI152NLin6Asym
g_fsl  = fmri_data(which('transcriptomic_gradients_MNI152NLin6Asym.nii.gz'));
% fmriprep / MNI152NLin2009cSym
g_fmp  = fmri_data(which('transcriptomic_gradients_MNI152NLin2009cSym.nii.gz'));
% CIFTI / 91k grayordinates
cii    = cifti_read(which('transcriptomic_gradients.dscalar.nii'));

Each 4D NIfTI holds three volumes — gradient 1, 2, 3 — in that order.

File inventory

File	Type	What it is
transcriptomic_gradients_MNI152NLin6Asym.nii.gz	NIfTI 4D	First 3 AHBA expression gradients in FSL/MNI152NLin6Asym space.
transcriptomic_gradients_MNI152NLin2009cSym.nii.gz	NIfTI 4D	Same gradients in fmriprep/MNI152NLin2009cSym space.
transcriptomic_gradients.dscalar.nii	CIFTI	Grayordinate version (registration fusion).
create_gradients.py	Python	Construction script, condensed from PennLINC's Vogel_PLS_Tx-Space.
README.md	Markdown	Methods write-up (B. Petre, 11/28/2024).
visualize_contents.m	MATLAB	Generates png_images/.

Citations

• Vogel JW, Alexander-Bloch AF, Wagstyl K, et al. (2024). Deciphering the functional specialization of whole-brain spatiomolecular gradients in the adult brain. PNAS 121:e2219137121. doi:10.1073/pnas.2219137121
• Hawrylycz MJ, Lein ES, Guillozet-Bongaarts AL, et al. (2012). An anatomically comprehensive atlas of the adult human brain transcriptome. Nature 489:391–399. doi:10.1038/nature11405
• Gorgolewski KJ, Fox AS, Chang L, et al. (2014). Tight fitting genes: finding relations between statistical maps and gene expression patterns. OHBM (poster + neurovault scripts).
• Markello RD, Arnatkeviciute A, Poline JB, Fulcher BD, Fornito A, Misic B (2021). Standardizing workflows in imaging transcriptomics with the abagen toolbox. eLife 10:e72129. doi:10.7554/eLife.72129