# generate re-ordering of coordinates based on Y-position
start = end = 0
data = np.zeros((len(coords)))
order = np.zeros(len(coords), dtype=int)
inds = np.arange(len(coords), dtype=int)
for i in np.asarray(info.groupby('hemisphere').count()['id']):
end += i
c = coords[start:end]
order[start:end] = inds[c[:, 1].argsort()[::-1] + start]
data[start:end] = np.arange(i)
start = i
np.put(data, order, data)
OPTS['vmax'] = len(data) - n_right
# plot original data and save
save_brainmap(data, FIGDIR / 'raw_surf.png', lh, rh, **OPTS)
spins = ['naive-nonpara', 'vazquez-rodriguez', 'vasa', 'hungarian', 'baum']
for sptype in spins:
spin = np.loadtxt(SPINDIR / name / sptype / f'{scale}_spins.csv',
delimiter=',',
dtype=int,
usecols=0)
plot = np.append(data, [np.nan])[spin] # accounts for 'baum' (-1 idx)
save_brainmap(plot, FIGDIR / f'{sptype}_surf.png', lh, rh, **OPTS)
# cornblath
spin = np.loadtxt(SPINDIR / 'vertex' / 'vazquez-rodriguez' /
'fsaverage5_spins.csv',
delimiter=',',
dtype=int,
# scatterplots comparing them to one another
for method, surrs in zip(METHODS, [burt2018, burt2020, moran]):
figdir = FIGDIR / name / method
figdir.mkdir(exist_ok=True, parents=True)
# brain maps of example brains
vmin, vmax = np.percentile(surrs, [2.5, 97.5])
for n, medial in enumerate(['medial', 'nomedial']):
fn = figdir / medial / f'{scale}.png'
if fn.exists():
continue
plotting.save_brainmap(surrs[n][:, 0],
fn,
annot.lh,
annot.rh,
subject_id='fsaverage5',
colormap='coolwarm',
colorbar=False,
vmin=vmin,
vmax=vmax,
views=['lat', 'med'])
# scatter plot of example brains
fig, ax = plt.subplots(1, 1)
ax.scatter(surrs[0][:, 0],
surrs[1][:, 0],
s=75,
edgecolor=np.array([60, 60, 60]) / 255,
facecolor=np.array([223, 121, 122]) / 255)
for side in ['right', 'top']:
ax.spines[side].set_visible(False)
ax.set(xlabel='with medial wall', ylabel='without medial wall')
OVERWRITE = True
OPTS = dict(subject_id='fsaverage5', views=['lat', 'med'], colormap='coolwarm')
if __name__ == "__main__":
FIGDIR.mkdir(parents=True, exist_ok=True)
parcellations = utils.get_cammoun_schaefer(data_dir=ROIDIR)
for alpha in simnulls.ALPHAS:
# create simulation with desired alpha (same seed as all others)
sim = spatial.create_surface_grf(alpha=float(alpha[-3:]),
seed=1234,
medial_val=np.nan)
# save plot of vertex-wise simulation
fn = FIGDIR / alpha / 'vertex' / 'fsaverage5.png'
if not fn.exists() or OVERWRITE:
save_brainmap(sim, fn, **OPTS)
# save plot of parcellated simulations (if desired)
if SAVE_ALL:
for parc, annots in parcellations.items():
for scale, annot in annots.items():
# parcellate simulation according
data = np.hstack([
utils.parcellate(np.split(sim, 2)[n], hemi, FSIGNORE)
for n, hemi in enumerate(annot)
])
# save plot of parcellated simulation
fn = FIGDIR / alpha / parc / f'{scale}.png'
if not fn.exists() or OVERWRITE:
save_brainmap(data,
fn,
diagonal=False, cbar=True,
cbar_label='correlation (r)',
imshow_kwargs=dict(vmin=-0.5, vmax=1, cmap=cmap))
# save resampling arrays
plot_save_heatmap(np.column_stack([range(len(data)), spins]),
fname='resamples', diagonal=True, cbar=False,
aspect='auto', cmap=cmap,
imshow_kwargs=dict(cmap=cmap))
# save sample brain maps
for term in terms:
# symmetrical colorbar
vmin, vmax = np.percentile(data[term], [2.5, 97.5])
lim = max(abs(vmin), abs(vmax))
save_brainmap(data[term], f'{term}_0.png', lh=lh, rh=rh,
colormap=cmap, vmin=-lim, vmax=lim)
# get spins, in case we want those, too
for n, spin in enumerate(spins.T, 1):
plot = np.asarray(data[term])[spin]
save_brainmap(plot, f'{term}_{n}.png', lh=lh, rh=rh,
colormap=cmap, vmin=-lim, vmax=lim)
# save example null distribution
null_dist = np.loadtxt(
NSDIR / PARC / 'nulls' / SPINTYPE / f'{SCALE}_nulls.csv'
)
fig, ax = plt.subplots(1, 1)
ax = sns.kdeplot(null_dist, color=cmap(0), linewidth=3, shade=True)
sns.despine(ax=ax, left=True)
ax.set(yticks=[], xlim=(0.4, 1.1), xticks=(0.5, 0.75, 1.0))
(x, y), = np.where(np.isin(data.columns, terms))
t1t2 = pd.read_csv(path, index_col=0) \
.assign(networks=RSN_AFILLIATION[PARC](SCALE).parcels) \
.drop(drop)
netmeans = t1t2.groupby('networks') \
.mean() \
.assign(network=YEO_CODES.keys())
spins = np.loadtxt(SPINDIR / PARC / SPINTYPE / f'{SCALE}_spins.csv',
delimiter=',', dtype='int32', usecols=range(3))
# get annotations
fetcher = getattr(nndata, f'fetch_{PARC.split("-")[1]}')
lh, rh = fetcher('fsaverage', data_dir=ROIDIR)[SCALE]
# plot original
data = np.asarray(t1t2['myelin'])
save_brainmap(data, 't1wt2w_0.png', lh=lh, rh=rh,
colormap='coolwarm', vmin=1.0, vmax=1.6)
for n, spin in enumerate(spins.T, 1):
save_brainmap(data[spin], f't1wt2w_{n}.png', lh=lh, rh=rh,
colormap='coolwarm', vmin=1.0, vmax=1.6)
null_dist = np.loadtxt(
HCPDIR / PARC / 'nulls' / 'yeo' / SPINTYPE / f'{SCALE}_nulls.csv',
delimiter=','
)
nulls = pd.DataFrame(dict(
network=np.repeat(list(YEO_CODES.keys()), len(null_dist)),
nulls=null_dist.flatten(order='F')
))
fig, ax = plt.subplots(1, 1, figsize=(4, 6))
ax = sns.boxplot(x='nulls', y='network', data=nulls, color='white',