Tera-MIND: Tera-scale mouse brain simulation via spatial mRNA-guided diffusion.

Code ↔ Paper

3 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 3 matches

[1] § Results › Tera-MIND accurately generated tera-scale mouse brain(s) by spatial gene expression ↔ utils/metrics.py, lines 201–215 · score 0.59 · peak signal, noise ratio, metrics, PSNR
[2] § Results › Tera-MIND achieved reproducible and robust results on three tera-scale mouse brains ↔ utils/vis_mba.py, lines 278–340 · score 0.56 · Nr4a2, Slc17a6, Slc17a7, Pathway, DOPA, GLUT
[3] § Results › Tera-MIND achieved reproducible and robust results on three tera-scale mouse brains ↔ utils/__init__.py, lines 59–95 · score 0.52 · Nr4a2, Slc17a6, Slc17a7, DOPA, GLUT

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

Python · 557 lines · 18 KB · MIT · 1 match

import torch
import pickle
import random
import numpy as np
import torch.nn.functional as F
import matplotlib.pyplot as plt
from cellpose import plot as cplt
import warnings
from torch.linalg import eigvals
def is_valid(pnm, gnm):
for p in pnm:
if p not in gnm:
return False
return True
def dt_region(hst, wst, hnm, wnm,
size=256, is_gt=True):
pad = size // 2
dt_lst = []
for ph in range(hnm):
hsz = ph * size
for pw in range(wnm):
wsz = pw * size
pnm = [hst + hsz, hst + size + hsz,
wst + wsz, wst + size + wsz]
if is_gt:
pnm += [hst - pad + hsz, hst + size + pad + hsz,
wst - pad + wsz, wst + size + pad + wsz]
pnm = '_'.join([str(p) for p in pnm])
dt_lst.append(pnm)
return dt_lst
def dt_sublst(gnm, hst=256, wst=256, hnm=16, wnm=16,
size=256, step=1, is_gt=True):
# for 1024 x 1024, hst=wst=512,
# hnm=412, wnm=284, step = 4
dt_lst= []
# Here, we assume no boundary issue
for pw in range(0, wnm, step):
wsz = pw * size
for ph in range(0, hnm, step):
hsz = ph * size
dt_reg = dt_region(hst+hsz, wst+wsz,
step, step, is_gt=is_gt)
if is_valid(dt_reg, gnm):
dt_lst.append(dt_reg)
return dt_lst
def _d_novel(sigma1, sigma2):
r"""
The core and more efficient impl of d_FID
Args:
sigma1: Covariance of one image collection
sigma2: Covariance of compared image collection
"""
eigval = eigvals(sigma1 @ sigma2)
eigval = eigval.real
eigval[eigval < 0] = 0
return 2 * eigval.sqrt().sum(-1)
def calc_d_fid(mu1, mu2, sigma1, sigma2):
r"""
The Function of d_FID calc
Args:
mu1: Mean of one image feat collection
mu2: Mean of compared image feat collection
sigma1: Covariance of one image feat collection
sigma2: Covariance of compared image feat collection
"""
mu1 = torch.atleast_1d(mu1)
mu2 = torch.atleast_1d(mu2)
sigma1 = torch.atleast_2d(sigma1)
sigma2 = torch.atleast_2d(sigma2)
assert mu1.shape == mu2.shape, \
'Training and test mean vectors have different lengths'
assert sigma1.shape == sigma2.shape, \
'Training and test covariances have different dimensions'
diff = mu1 - mu2
fid_easy = diff.dot(diff) + torch.trace(sigma1) + torch.trace(sigma2)
fid_hard = _d_novel(sigma1, sigma2)
fid = fid_easy - fid_hard
return fid
def calc_d_fid3(mu1, mu2, sigma1, sigma2):
mu1 = torch.atleast_2d(mu1)
mu2 = torch.atleast_2d(mu2)
sigma1 = torch.atleast_3d(sigma1)
sigma2 = torch.atleast_3d(sigma2)
assert mu1.shape == mu2.shape, \
'Training and test mean vectors have different lengths'
assert sigma1.shape == sigma2.shape, \
'Training and test covariances have different dimensions'
dif = mu1 - mu2
fid_easy = (dif ** 2).sum(-1) + torch.vmap(torch.trace)(sigma1) + torch.vmap(torch.trace)(sigma2)
fid_hard = _d_novel(sigma1, sigma2)
fid = fid_easy - fid_hard
return fid
def calc_mean_var(mu, scm, tot):
mu = mu / tot[..., None]
scm = scm / tot[..., None, None]
sigma = scm - mu.unsqueeze(-1) @ mu.unsqueeze(-2)
return mu, sigma
def calc_slc_all_fid(m_r, s_r, tot_r,
m_g, s_g, tot_g, is_str=False):
m_r0, s_r0 = calc_mean_var(m_r, s_r, tot_r)
m_g0, s_g0 = calc_mean_var(m_g, s_g, tot_g)
fid0 = calc_d_fid3(m_r0, m_g0, s_r0, s_g0)
m_r1, s_r1 = calc_mean_var(m_r.sum(0)[None], s_r.sum(0)[None], tot_r.sum(0)[None])
m_g1, s_g1 = calc_mean_var(m_g.sum(0)[None], s_g.sum(0)[None], tot_g.sum(0)[None])
fid1 = calc_d_fid3(m_r1, m_g1, s_r1, s_g1)
fid = torch.cat((fid0, fid1)).cpu().numpy()
if is_str:
return [f'{fd:.2f}' for fd in fid]
return fid
def calc_mean_std_msk(stat, dims):
msk = stat!=0
avg = (stat*msk).sum(dims)/msk.sum(dims)
# Here, we always assume (batch, z) for stat
var = (((stat - avg)**2)*msk).sum(dims)/msk.sum(dims)
return avg, var.sqrt()
def calc_slc_all_1d(stat, is_str=False):
avg0, std0 = calc_mean_std_msk(stat, 0)
avg1, std1 = calc_mean_std_msk(stat, (0, 1))
avg = torch.cat((avg0, avg1[None]))
std = torch.cat((std0, std1[None]))
if is_str:
return [f'{a:.2f} {s:.2f}' for a, s in zip(avg, std)]
return avg, std
def calc_cellpose(img, mod, met,
pth, roi, debug):
im = img.astype('float') / 255.
im_lst = np.split(im[:, 0], im.shape[0], 0)
im_lst = [(v, i[0]) for v, i in enumerate(im_lst) if (img[v, 0] != 0).any()]
if not im_lst:
print(f'{roi} all black, ignore.')
return
vl_lst, mk_lst = zip(*im_lst)
masks, flows, styles, diams = mod.eval(
list(mk_lst), diameter=None,
normalize=False, channels=[0, 0])
expr = im.mean((-1, -2))
out = [torch.zeros(3, im.shape[0]),
# Estimate the maximum num of cells
torch.zeros(512, im.shape[0]),
vl_lst]
for mid, msk in zip(vl_lst, masks):
cnt = np.unique(msk, return_counts=True)[1]
cnt = cnt[1:]
out[0][0, mid] = len(cnt)
out[0][1:, mid] = torch.FloatTensor(expr[mid])
out[1][:len(cnt), mid] = torch.FloatTensor(cnt)
mlen = int(out[0][0].max())
assert mlen <= out[1].shape[0]
out[1] = out[1][:mlen]
met['nstat'].append(out[0])
met['narea'].append(out[1])
met['valid'].append(out[2])
if debug:
print(roi, vl_lst, expr.shape, out[1].shape)
for mid in range(im.shape[0]):
if mid not in vl_lst:
if (expr[mid] != 0).any():
print(mid, expr[mid])
_s = random.choice(range(len(vl_lst)))
fig = plt.figure(figsize=(12, 5))
cplt.show_segmentation(fig, img[vl_lst[_s], 0], masks[_s],
flows[_s][0], channels=[0, 0])
plt.tight_layout()
plt.savefig(str(pth / f'{roi}_{vl_lst[_s]}.png'),
bbox_inches='tight', dpi=200)
plt.close()
class PSNR(torch.nn.Module):
"""Peak Signal to Noise Ratio
img1 and img2 have range [0, 255]"""
def __init__(self, mval=255.):
super(PSNR, self).__init__()
self.mval = mval
def forward(self, img1, img2):
if len(img1.shape) == 3:
dim = [1, 2]
elif len(img1.shape) == 4:
dim = [1, 2, 3]
mse = torch.mean((img1 - img2) ** 2, dim=dim)
return 20 * torch.log10(self.mval / torch.sqrt(mse))
def _fspecial_gauss_1d(size, sigma):
r"""Create 1-D gauss kernel
Args:
size (int): the size of gauss kernel
sigma (float): sigma of normal distribution
Returns:
torch.Tensor: 1D kernel (1 x 1 x size)
"""
coords = torch.arange(size, dtype=torch.float)
coords -= size // 2
g = torch.exp(-(coords ** 2) / (2 * sigma ** 2))
g /= g.sum()
return g.unsqueeze(0).unsqueeze(0)
def gaussian_filter(input, win):
r""" Blur input with 1-D kernel
Args:
input (torch.Tensor): a batch of tensors to be blurred
window (torch.Tensor): 1-D gauss kernel
Returns:
torch.Tensor: blurred tensors
"""
assert all([ws == 1 for ws in win.shape[1:-1]]), win.shape
if len(input.shape) == 4:
conv = F.conv2d
elif len(input.shape) == 5:
conv = F.conv3d
else:
raise NotImplementedError(input.shape)
C = input.shape[1]
out = input
for i, s in enumerate(input.shape[2:]):
if s >= win.shape[-1]:
out = conv(out, weight=win.transpose(2 + i, -1), stride=1, padding=0, groups=C)
else:
warnings.warn(
f"Skipping Gaussian Smoothing at dimension 2+{i} for input: {input.shape} and win size: {win.shape[-1]}"
)
return out
def _ssim(X, Y, data_range, win, size_average=True, K=(0.01, 0.03)):
r""" Calculate ssim index for X and Y
Args:
X (torch.Tensor): images
Y (torch.Tensor): images
win (torch.Tensor): 1-D gauss kernel
data_range (float or int, optional): value range of input images. (usually 1.0 or 255)
size_average (bool, optional): if size_average=True, ssim of all images will be averaged as a scalar
Returns:
torch.Tensor: ssim results.
"""
K1, K2 = K
# batch, channel, [depth,] height, width = X.shape
compensation = 1.0
C1 = (K1 * data_range) ** 2
C2 = (K2 * data_range) ** 2
win = win.to(X.device, dtype=X.dtype)
mu1 = gaussian_filter(X, win)
mu2 = gaussian_filter(Y, win)
mu1_sq = mu1.pow(2)
mu2_sq = mu2.pow(2)
mu1_mu2 = mu1 * mu2
sigma1_sq = compensation * (gaussian_filter(X * X, win) - mu1_sq)
sigma2_sq = compensation * (gaussian_filter(Y * Y, win) - mu2_sq)
sigma12 = compensation * (gaussian_filter(X * Y, win) - mu1_mu2)
cs_map = (2 * sigma12 + C2) / (sigma1_sq + sigma2_sq + C2) # set alpha=beta=gamma=1
ssim_map = ((2 * mu1_mu2 + C1) / (mu1_sq + mu2_sq + C1)) * cs_map
ssim_per_channel = torch.flatten(ssim_map, 2).mean(-1)
cs = torch.flatten(cs_map, 2).mean(-1)
return ssim_per_channel, cs
def ssim(
X,
Y,
data_range=255,
size_average=True,
win_size=11,
win_sigma=1.5,
win=None,
K=(0.01, 0.03),
nonnegative_ssim=False,
):
r""" interface of ssim
Args:
X (torch.Tensor): a batch of images, (N,C,H,W)
Y (torch.Tensor): a batch of images, (N,C,H,W)
data_range (float or int, optional): value range of input images. (usually 1.0 or 255)
size_average (bool, optional): if size_average=True, ssim of all images will be averaged as a scalar
win_size: (int, optional): the size of gauss kernel
win_sigma: (float, optional): sigma of normal distribution
win (torch.Tensor, optional): 1-D gauss kernel. if None, a new kernel will be created according to win_size and win_sigma
K (list or tuple, optional): scalar constants (K1, K2). Try a larger K2 constant (e.g. 0.4) if you get a negative or NaN results.
nonnegative_ssim (bool, optional): force the ssim response to be nonnegative with relu
Returns:
torch.Tensor: ssim results
"""
if not X.shape == Y.shape:
raise ValueError(f"Input images should have the same dimensions, but got {X.shape} and {Y.shape}.")
for d in range(len(X.shape) - 1, 1, -1):
X = X.squeeze(dim=d)
Y = Y.squeeze(dim=d)
if len(X.shape) not in (4, 5):
raise ValueError(f"Input images should be 4-d or 5-d tensors, but got {X.shape}")
if not X.type() == Y.type():
raise ValueError(f"Input images should have the same dtype, but got {X.type()} and {Y.type()}.")
if win is not None: # set win_size
win_size = win.shape[-1]
if not (win_size % 2 == 1):
raise ValueError("Window size should be odd.")
if win is None:
win = _fspecial_gauss_1d(win_size, win_sigma)
win = win.repeat([X.shape[1]] + [1] * (len(X.shape) - 1))
ssim_per_channel, cs = _ssim(X, Y, data_range=data_range, win=win, size_average=False, K=K)
if nonnegative_ssim:
ssim_per_channel = torch.relu(ssim_per_channel)
if size_average:
return ssim_per_channel.mean()
else:
return ssim_per_channel.mean(1)
def ms_ssim(
X, Y, data_range=255, size_average=True, win_size=11, win_sigma=1.5, win=None, weights=None, K=(0.01, 0.03)
):
r""" interface of ms-ssim
Args:
X (torch.Tensor): a batch of images, (N,C,[T,]H,W)
Y (torch.Tensor): a batch of images, (N,C,[T,]H,W)
data_range (float or int, optional): value range of input images. (usually 1.0 or 255)
size_average (bool, optional): if size_average=True, ssim of all images will be averaged as a scalar
win_size: (int, optional): the size of gauss kernel
win_sigma: (float, optional): sigma of normal distribution
win (torch.Tensor, optional): 1-D gauss kernel. if None, a new kernel will be created according to win_size and win_sigma
weights (list, optional): weights for different levels
K (list or tuple, optional): scalar constants (K1, K2). Try a larger K2 constant (e.g. 0.4) if you get a negative or NaN results.
Returns:
torch.Tensor: ms-ssim results
"""
if not X.shape == Y.shape:
raise ValueError(f"Input images should have the same dimensions, but got {X.shape} and {Y.shape}.")
for d in range(len(X.shape) - 1, 1, -1):
X = X.squeeze(dim=d)
Y = Y.squeeze(dim=d)
if not X.type() == Y.type():
raise ValueError(f"Input images should have the same dtype, but got {X.type()} and {Y.type()}.")
if len(X.shape) == 4:
avg_pool = F.avg_pool2d
elif len(X.shape) == 5:
avg_pool = F.avg_pool3d
else:
raise ValueError(f"Input images should be 4-d or 5-d tensors, but got {X.shape}")
if win is not None: # set win_size
win_size = win.shape[-1]
if not (win_size % 2 == 1):
raise ValueError("Window size should be odd.")
smaller_side = min(X.shape[-2:])
assert smaller_side > (win_size - 1) * (
2 ** 4
), "Image size should be larger than %d due to the 4 downsamplings in ms-ssim" % ((win_size - 1) * (2 ** 4))
if weights is None:
weights = [0.0448, 0.2856, 0.3001, 0.2363, 0.1333]
weights = X.new_tensor(weights)
if win is None:
win = _fspecial_gauss_1d(win_size, win_sigma)
win = win.repeat([X.shape[1]] + [1] * (len(X.shape) - 1))
levels = weights.shape[0]
mcs = []
for i in range(levels):
ssim_per_channel, cs = _ssim(X, Y, win=win, data_range=data_range, size_average=False, K=K)
if i < levels - 1:
mcs.append(torch.relu(cs))
padding = [s % 2 for s in X.shape[2:]]
X = avg_pool(X, kernel_size=2, padding=padding)
Y = avg_pool(Y, kernel_size=2, padding=padding)
ssim_per_channel = torch.relu(ssim_per_channel) # (batch, channel)
mcs_and_ssim = torch.stack(mcs + [ssim_per_channel], dim=0) # (level, batch, channel)
ms_ssim_val = torch.prod(mcs_and_ssim ** weights.view(-1, 1, 1), dim=0)
if size_average:
return ms_ssim_val.mean()
else:
return ms_ssim_val.mean(1)
class SSIM(torch.nn.Module):
def __init__(
self,
data_range=255,
size_average=True,
win_size=11,
win_sigma=1.5,
channel=3,
spatial_dims=2,
K=(0.01, 0.03),
nonnegative_ssim=False,
unsqueeze=False
):
r""" class for ssim
Args:
data_range (float or int, optional): value range of input images. (usually 1.0 or 255)
size_average (bool, optional): if size_average=True, ssim of all images will be averaged as a scalar
win_size: (int, optional): the size of gauss kernel
win_sigma: (float, optional): sigma of normal distribution
channel (int, optional): input channels (default: 3)
K (list or tuple, optional): scalar constants (K1, K2). Try a larger K2 constant (e.g. 0.4) if you get a negative or NaN results.
nonnegative_ssim (bool, optional): force the ssim response to be nonnegative with relu.
"""
super(SSIM, self).__init__()
self.win_size = win_size
self.win = _fspecial_gauss_1d(win_size, win_sigma).repeat([channel, 1] + [1] * spatial_dims)
self.size_average = size_average
self.data_range = data_range
self.K = K
self.nonnegative_ssim = nonnegative_ssim
self.unsqueeze = unsqueeze
def forward(self, X, Y):
if self.unsqueeze:
# this suggests that the input
# has only 3 channels [N, H, W]
# then unsqueeze the C channel
assert len(X.shape) == len(Y.shape) == 3
X = X.unsqueeze(1)
Y = Y.unsqueeze(1)
return ssim(
X,
Y,
data_range=self.data_range,
size_average=self.size_average,
win=self.win,
K=self.K,
nonnegative_ssim=self.nonnegative_ssim,
)
class MS_SSIM(torch.nn.Module):
def __init__(
self,
data_range=255,
size_average=True,
win_size=11,
win_sigma=1.5,
channel=3,
spatial_dims=2,
weights=None,
K=(0.01, 0.03),
unsqueeze=False
):
r""" class for ms-ssim
Args:
data_range (float or int, optional): value range of input images. (usually 1.0 or 255)
size_average (bool, optional): if size_average=True, ssim of all images will be averaged as a scalar
win_size: (int, optional): the size of gauss kernel
win_sigma: (float, optional): sigma of normal distribution
channel (int, optional): input channels (default: 3)
weights (list, optional): weights for different levels
K (list or tuple, optional): scalar constants (K1, K2). Try a larger K2 constant (e.g. 0.4) if you get a negative or NaN results.
"""
super(MS_SSIM, self).__init__()
self.win_size = win_size
self.win = _fspecial_gauss_1d(win_size, win_sigma).repeat([channel, 1] + [1] * spatial_dims)
self.win_gray = _fspecial_gauss_1d(win_size, win_sigma).repeat([1, 1] + [1] * spatial_dims)
self.size_average = size_average
self.data_range = data_range
self.weights = weights
self.K = K
self.unsqueeze = unsqueeze
def forward(self, X, Y):
is_gray = False
if len(X.shape) == 3:
X, Y = X[:, None], Y[:, None]
is_gray = True
return ms_ssim(
X,
Y,
data_range=self.data_range,
size_average=self.size_average,
win=self.win_gray if is_gray else self.win,
weights=self.weights,
K=self.K,
)
if __name__ == '__main__':
import torch
# mu = torch.rand(3, 50)
# scm = torch.rand(3, 50, 50)
# tot = torch.rand(3)
# a = mu.clone()
# m0, s0 = calc_mean_var(mu, scm, tot)
# print((a == mu).all())
# ssim_ = SSIM(channel=2, size_average=False)
# a = torch.rand(23, 2, 125,125)
# b = torch.rand(23, 2, 125,125)
# c = ssim_(a, b)
# print(c.shape)

metrics.py at commit 083747d, under MIT · at the source

Overview

Authors: Jiqing Wu¹, Ingrid Berg², Yawei Li^3,4, Ender Konukoglu³, Viktor H Koelzer^1,5

ORCID iDs: Jiqing Wu

Institutions: University of Basel (Switzerland); University of Zurich (Switzerland); University Hospital Zurich (Switzerland); ETH Zurich (Switzerland); Integrated Systems Laboratory (Switzerland); Computer Vision Laboratory (Switzerland); University Hospital of Basel (Switzerland)

Journal: iScience, volume 29, issue 7, article 116355

Dates: received 10 October 2025; accepted 6 April 2026; published online 8 July 2026

Type: Research article · Language: English

License: CC BY

Identifiers: DOI 10.1016/j.isci.2026.116355 · PMID 42491506 · PMCID PMC13378011 · OpenAlex W4415338870

Open access: gold, a free copy (OpenAlex)

Preprint: arxiv:2503.01220

Status: code verified

Categories: genetics / omics (modality), mouse (organism), cellular / molecular (subfield)

Methods: Machine learning

Keywords: Neuroscience, Biocomputational method, Transcriptomics

Topic: RNA Research and Splicing (Molecular Biology, Biochemistry, Genetics and Molecular Biology), according to OpenAlex

Funding: Universitätsspital Basel

Citations: not cited yet (Europe PMC); 52 references in the paper

Abstract

Holistic 3D modeling of molecularly defined brain structures is crucial for understanding complex brain functions. Using emerging tissue profiling technologies, researchers charted comprehensive atlases of mammalian brain with sub-cellular resolution and spatially resolved transcriptomic data. However, these tera-scale volumetric atlases pose computational challenges for modeling intricate brain structures within the native spatial context. We propose Tera-MIND, a novel generative framework capable of simulating Tera-scale mouse brains in 3D using a patch-based and boundary-aware diffusion model. Taking spatial gene expression as conditional input, we generate virtual mouse brains with comprehensive cellular morphological detail at teravoxel scale. Through the lens of 3D gene-gene self-attention, we identify spatial molecular interactions for key transcriptomic pathways, including glutamatergic and dopaminergic neuronal systems. Lastly, we showcase the translational applicability of Tera-MIND on previously unseen human brain samples. Tera-MIND offers an efficient generative modeling of whole virtual organisms, paving the way for integrative applications in biomedical research.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repositories

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

Zenodo 14826874

License: MIT

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Size: 0 files, 0 scripts

Software Heritage: not checked

Found in: “Data and code availability”

Not found: README, license file, CITATION.cff, environment file, tests, continuous integration, documentation

Availability: 1 check, the latest on 27 September 2026: the link answers (HTTP 200)

27 September 2026: the link answers (HTTP 200)

At the source:

CTPLab/Tera-MIND

License: MIT

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Commit: 083747d0ddc3d17ccbb31c5556b9f319d375852f, 24 June 2025

Languages: Python (32), Shell (7)

Size: 52 files, 39 scripts

Software Heritage: not archived

Found in: “Data and code availability”

Holds: README, license file, environment (environment.yml)

Not found: CITATION.cff, tests, continuous integration, documentation

Tools: PyTorch (23 files), NumPy (15 files), pandas (7 files), Pillow (6 files), Matplotlib (4 files), Cellpose (3 files), seaborn (3 files), OpenCV (2 files), PyTorch Lightning (1 file), SciPy (1 file)

Availability: 1 check, the latest on 27 September 2026: the link answers

27 September 2026: the link answers

41 files

config.py, Python, 326 lines
config_base.py, Python, 72 lines
config_parm.py, Python, 59 lines
diffusion/__init__.py, Python, 6 lines
diffusion/base.py, Python, 730 lines
diffusion/diffusion.py, Python, 161 lines
diffusion/resample.py, Python, 63 lines
experiment.py, Python, 491 lines
infer_DOPA.sh, Shell, 5 lines
infer_GLUT.sh, Shell, 5 lines
infer_attn.py, Python, 86 lines
infer_brn.py, Python, 156 lines
infer_brn.sh, Shell, 6 lines
model/MBAblocks.py, Python, 665 lines
model/__init__.py, Python, 5 lines
model/blocks.py, Python, 632 lines
model/nn.py, Python, 262 lines
model/unet_attn.py, Python, 267 lines
model/unet_ours.py, Python, 476 lines
model/unet_patch_dm.py, Python, 577 lines
model/unet_sinf.py, Python, 270 lines
test_DOPA.sh, Shell, 4 lines
test_GLUT.sh, Shell, 4 lines
test_attn.py, Python, 589 lines
test_brn.py, Python, 351 lines
test_brn.sh, Shell, 4 lines
train.py, Python, 45 lines
train.sh, Shell, 2 lines
utils/MBADataset.py, Python, 287 lines
utils/MBADataset_tst.py, Python, 203 lines
utils/__init__.py, Python, 95 lines, 1 match
utils/choices.py, Python, 180 lines
utils/dataset_util.py, Python, 77 lines
utils/dist_utils.py, Python, 42 lines
utils/feat_mba.py, Python, 564 lines
utils/inft.py, Python, 476 lines
utils/inft_mba.py, Python, 262 lines
utils/metrics.py, Python, 557 lines, 1 match
utils/vis_mba.py, Python, 514 lines, 1 match
LICENSE, License, 21 lines
README.md, Text, 143 lines

At the source: github.com/CTPLab/Tera-MIND

musikisomorphie/tera-mind

License: none: the authors keep all their rights

State: the link is dead, verified on 27 September 2026

Evidence: found in the paper

Software Heritage: not archived

Found in: the Zenodo archive record

Not found: README, license file, CITATION.cff, environment file, tests, continuous integration, documentation

Availability: 1 check, the latest on 27 September 2026: the link is dead

27 September 2026: the link is dead

At the source: github.com/musikisomorphie/Tera-MIND/tree/v1.0.0

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

3 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
39 scripts, each with its path and the digest of its content;
3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data and code availability

The original mouse brain data were deposited at brain image library under the database identifier ace-den-fix. The processed mouse brain data were deposited at brain image library under the database identifier ace-lot-now and are publicly available as of the date of publication. The human brain data are available via https://www.10xgenomics.com/datasets/xenium-human-brain-preview-data-1-standard.

All original code has been deposited at Zenodo under the https://doi.org/10.5281/zenodo.14826874 or https://github.com/CTPLab/Tera-MIND.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 2, 28 September 2026

Authors: added Jiqing Wu (0000-0002-6898-8698); removed Jiqing Wu

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 5 authors, 3 keywords, 1 funder, 36 references.

Cite

This paper

Wu, J., Berg, I., Li, Y., Konukoglu, E., & Koelzer, V. H. (2026). Tera-MIND: Tera-scale mouse brain simulation via spatial mRNA-guided diffusion. iScience, 29(7), 116355. https://doi.org/10.1016/j.isci.2026.116355

BibTeX

@article{wu2026tera, author = {Wu, Jiqing and Berg, Ingrid and Li, Yawei and Konukoglu, Ender and Koelzer, Viktor H}, title = {{Tera-MIND: Tera-scale mouse brain simulation via spatial mRNA-guided diffusion}}, journal = {iScience}, year = {2026}, month = jul, volume = {29}, number = {7}, pages = {116355}, publisher = {Elsevier}, issn = {2589-0042}, doi = {10.1016/j.isci.2026.116355}, url = {https://doi.org/10.1016/j.isci.2026.116355}, pmid = {42491506}, pmcid = {PMC13378011} }

RIS

TY - JOUR AU - Wu, Jiqing AU - Berg, Ingrid AU - Li, Yawei AU - Konukoglu, Ender AU - Koelzer, Viktor H TI - Tera-MIND: Tera-scale mouse brain simulation via spatial mRNA-guided diffusion T2 - iScience J2 - iScience PY - 2026 DA - 2026/07/08 VL - 29 IS - 7 SP - 116355 SN - 2589-0042 PB - Elsevier DO - 10.1016/j.isci.2026.116355 UR - https://doi.org/10.1016/j.isci.2026.116355 LA - en ER -

CSL-JSON

{ "id": "10.1016/j.isci.2026.116355", "type": "article-journal", "title": "Tera-MIND: Tera-scale mouse brain simulation via spatial mRNA-guided diffusion", "container-title": "iScience", "author": [ { "family": "Wu", "given": "Jiqing" }, { "family": "Berg", "given": "Ingrid" }, { "family": "Li", "given": "Yawei" }, { "family": "Konukoglu", "given": "Ender" }, { "family": "Koelzer", "given": "Viktor H" } ], "container-title-short": "iScience", "volume": "29", "issue": "7", "page": "116355", "DOI": "10.1016/j.isci.2026.116355", "PMID": "42491506", "PMCID": "PMC13378011", "ISSN": "2589-0042", "publisher": "Elsevier", "URL": "https://doi.org/10.1016/j.isci.2026.116355", "language": "en", "issued": { "date-parts": [ [ 2026, 7, 8 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1038/s41467-026-71759-4 [code]: CellNiche represents cellular microenvironments in atlas-scale spatial omics data with contrastive learning.
Journal: Nature communications
In common: PyTorch, seaborn, pandas, 3 other tools, mouse, cellular / molecular, 6 references
[2] doi:10.1371/journal.pcbi.1014571 [code]: SynAPSeg: A novel dataset and image analysis framework for deep learning-based synapse detection and quantification.
Journal: PLoS computational biology
In common: Cellpose, OpenCV, Pillow, 6 other tools, 2 references
[3] doi:10.1038/s42003-026-10957-8 [code]: Brain defence by the extracellular matrix protein Cochlin.
Journal: Communications biology
In common: PyTorch Lightning, OpenCV, Pillow, 6 other tools, mouse, cellular / molecular
[4] doi:10.1038/s41593-026-02293-1 [code]: Optics-free spatial genomics for mapping mammalian brain aging by IRISeq.
Journal: Nature neuroscience
In common: seaborn, pandas, SciPy, 2 other tools, genetics / omics, mouse, cellular / molecular, 4 references
[5] doi:10.1038/s41586-026-10323-y [code]: Genetically encoded assembly recorder temporally resolves cellular history.
Journal: Nature
In common: Cellpose, OpenCV, Pillow, 5 other tools, mouse, cellular / molecular
[6] doi:10.1038/s41467-026-72130-3 [code]: Retinoic acid drives cell fate specification, maturation and retinal regionality in human retinal organoids.
Journal: Nature communications
In common: Cellpose, OpenCV, Pillow, 4 other tools, 1 reference
[7] doi:10.1186/s12859-026-06490-4 [code]: Tissueformer: extending single-cell foundation models to predict population-level phenotypes.
Journal: BMC bioinformatics
In common: PyTorch, seaborn, pandas, 3 other tools, genetics / omics, mouse, cellular / molecular, 3 references
[8] doi:10.1186/s12864-026-12965-8 [code]: Systematic evaluation of single-cell foundation model interpretability: attention-derived edge scores add no incremental value over gene-level features for perturbation-target prediction.
Journal: BMC genomics
In common: PyTorch, seaborn, pandas, 3 other tools, cellular / molecular, 3 references
[9] doi:10.3389/frai.2026.1841639 [code]: Label tree semantic losses for rich multi-class medical image segmentation.
Journal: Frontiers in artificial intelligence
In common: PyTorch Lightning, OpenCV, Pillow, 6 other tools
[10] doi:10.1162/imag.a.1207 [code]: Investigating the temporal dynamics and modeling of mid-level feature representations in humans.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: PyTorch Lightning, OpenCV, Pillow, 6 other tools

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.