A critical initialization for biological neural networks.
10 blocks of code here have a permanent identifier (oscr:b:…), listed with the matches below: citable for good, with a page of its own. Select a match and press Cite to copy it. What a block identifier is.
The 13 matches
- [1] § Methods › Data acquisition › Neuropixel recordings and processing ↔ fig2/analysis.py, lines 269 to 325 · score 0.85 · hippocampal formation, subcortical areas, firing rate, visual cortex, cortices, striatum · block
oscr:b:99E64DC1RMK4Y-H· Report a mismatch - [2] § Intrinsic structure in neural recordings ↔ fig2/fig2.py, lines 10 to 72 · score 0.82 · 2385–10344 regions, 2961–8566 ROIs, 1716–2914, 2385 regions, 2961 ROIs, Neuropixels · block
oscr:b:NS9M9FN4SWWG2-Z· Report a mismatch - [3] § Intrinsic structure in neural recordings ↔ fig2/fig2.py, lines 10 to 72 · score 0.82 · 2385–10344 regions, 2961–8566 ROIs, 1716–2914, 2385 regions, 2961 ROIs, Neuropixels · block
oscr:b:2WAX6YQGTPV9V-T· Report a mismatch - [4] § Computations with symmetric dynamics ↔ fig5/fig5.ipynb, lines 61 to 163 · score 0.74 · zero shot working, persistent activity, working memory, symmetric dynamics, subspace, training · Report a mismatch
- [5] § Computations with symmetric dynamics ↔ fig5/fig5.ipynb, lines 61 to 163 · score 0.69 · zero shot working, working memory, symmetric dynamics, binary, training, models · Report a mismatch
- [6] § Methods › Data analysis › Estimating rotational components from data ↔ fig3/other_datasets.py, lines 405 to 474 · score 0.65 · linear track, virtual reality, visual cortex, reward, maze, PSTHs · block
oscr:b:8ZG0K75A87K8R-C· Report a mismatch - [7] § Methods › Data analysis › Estimating rotational components from data ↔ fig3/other_datasets.py, lines 405 to 474 · score 0.65 · linear track, virtual reality, visual cortex, reward, maze, PSTHs · block
oscr:b:8J65P2RPSXJHM-W· Report a mismatch - [8] § Methods ↔ fig_utils.py, lines 1 to 27 · score 0.62 · Howard Hughes Medical · block
oscr:b:CJRNHRW5A1QN0-2· Report a mismatch - [9] § Methods › Simulations of dynamical systems › Sparse/varied connectivity ↔ fig4/sparse_clustered_local_sims.py, lines 136 to 193 · score 0.58 · exponential decay, locally connected, Bernoulli, distances, global, Sparse · block
oscr:b:Q1H1AY3WB3J3H-E· Report a mismatch - [10] § Methods › Simulations of dynamical systems ↔ simulations.py, lines 241 to 319 · score 0.54 · shot noise, relu, SNR, Poisson, decay, simulate · block
oscr:b:KHG73G56PJD3N-P· Report a mismatch - [11] § Intrinsic structure in neural recordings ↔ fig2/fig2.py, lines 413 to 470 · score 0.54 · brainwide ephys, power law exponents, duration, SVCA2, CA1, neurons · block
oscr:b:79BXW3KTM0KGF-Q· Report a mismatch - [12] § Intrinsic structure in neural recordings ↔ fig2/fig2.py, lines 413 to 470 · score 0.54 · brainwide ephys, power law exponents, duration, SVCA2, CA1, neurons · block
oscr:b:H5BAS9EAF629R-U· Report a mismatch - [13] § Methods › Simulations of dynamical systems › Sparse/varied connectivity ↔ fig4/sparse_clustered_local_sims.py, lines 136 to 193 · score 0.53 · global connection, local connection, Bernoulli, clustered, Sparse, symmetric · block
oscr:b:Q1H1AY3WB3J3H-E· Report a mismatch
Paper
Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC
The paper is loaded when this pane is shown.
The authors' code
Python · 973 lines · 43 KB · GPL-3.0 · 2 matches
- from matplotlib.patches import Ellipse
- from fig_utils import *
- from scipy.stats import zscore
- from powerlaw import fit_powerlaw_exp
- from scipy.stats import ttest_rel
- from matplotlib import gridspec
- from scipy.interpolate import interp1d
- def fig2(dat):
- areas_all = dat["areas_all"]
- evals_all = dat["evals_svca2_all"]
- evals_shuff_all = dat["evals_shuff_all"]
- evals_run = dat["evals_run_all"]
- fig = plt.figure(figsize=(14, 7), dpi=300)
- yratio = 14/7
- grid = plt.GridSpec(8, 7, wspace=0.4, hspace=0.2, figure=fig,
- bottom=0.04, top=0.94, left=0.02, right=1.)
- titles = ["cortical 2P imaging", "CA1 2P imaging",
- "brainwide Neuropixels"]
- sneur = ["2,385 - 10,344 ROIs", "2,961 - 8,566 ROIs", "1,716 - 2,914 units"]
- colors = dcolors[:3].copy()
- n_dset = len(areas_all)
- ids = np.zeros(n_dset, "int")
- for d in range(len(areas)):
- ids[np.array(areas_all)==areas[d]] = d
- ls = ["-", "-", "-", "--", "-."]
- print(areas_all, ids)
- dy = [0, 2, 1]
- alphas = np.zeros(n_dset)
- alphas_shuff = np.zeros(n_dset)
- ymax = 500
- norms = np.zeros(n_dset)
- if "imexs" in dat:
- # example images/traces (extracted from suite2p folders from example data)
- # (if available make example panels)
- imexs = dat["imexs"]
- masksexs = dat["masksexs"]
- for d in range(3):
- ax = plt.subplot(grid[:2, d])
- pos = ax.get_position().bounds
- ax.set_position([pos[0]+0.04*d + 0.01, pos[1], pos[2]*(1.3 +0.4*(d<2)), pos[3]*1.])
- if d==2:
- im = plt.imread("allenprobes.png")
- #print(im.shape, imexs[1].shape)
- ax.imshow(im[58:-100])
- else:
- ax.imshow(imexs[d], vmin=0, vmax=0.9,
- cmap="gray", aspect=0.75/0.5 if d==1 else 1)
- masks0 = masksexs[d].copy()
- yo, xo = np.nonzero(masks0[:,:,-1]>0)
- masks0[yo,xo,-1] = 0.25
- ax.imshow(masks0, aspect=0.75/0.5 if d==1 else 1)
- if d==0:
- ax.set_ylim([380, 380+90])
- ax.set_xlim(50, 50 + 130)
- elif d==1:
- #ax.set_ylim([350, 430])
- #ax.set_xlim([100, 100 + 80*0.75/0.5])
- ax.set_ylim([50, 50+90])
- ax.set_xlim([250, 250+130*0.75/0.5])
- ax.set_title(titles[d], color=colors[d], fontstyle="italic",
- y=1.075, loc="left", x=-0.05)#, fontweight="bold")
- ax.text(0.5, 1.025, sneur[d], transform=ax.transAxes,
- fontsize="small", ha="center")
- ax.axis("off")
- il = d
- transl = mtransforms.ScaledTranslation(-25 / 72, 12 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- for j in range(2):
- Xexs = dat["Xexs"] if j==0 else dat["Xemb_ex"]
- Xexs = [Xexs[0], Xexs[1]] if j==0 else Xexs
- if j==0:
- titles = ['Rastermap of 2p data - V1', 'CA1']
- tmins = [5800, 42000//7+200] # 5000
- tlen = 22*60*3//7
- else:
- titles = ['Simulated 2p data - max eigenvalue = 0.998', 'max eigenvalue = 0.975']
- tmins = [5960*2, 0]
- tlen = 22*60*3//7
- for d in range(2):
- ax = plt.subplot(grid[2+3*j:5+3*j, 2*d:2*d+2])
- pos = ax.get_position().bounds
- ax.set_position([pos[0]-d*0.015, pos[1]-j*0.02, pos[2]+0.005, pos[3]*0.9])
- #Xe = zscore(Xexs[d][:, tmins[d]:tmins[d]+tlen].copy(), axis=1)
- im = ax.imshow(zscore(Xexs[d][:, tmins[d]:tmins[d]+tlen], axis=1),
- aspect="auto", cmap="gray_r", vmin=0., vmax=1.5)
- ax.plot([0, 22*30/7], -Xexs[d].shape[0]*0.04*np.ones(2), color="k", lw=1.5)
- ax.plot(-0.025*tlen*np.ones(2), [0, 50], color="k", lw=1.5)
- if d==0:
- if j==0:
- ax.text((22*30/2)/7, -50, "30 sec.", ha="center", va="center", fontsize="small")
- ax.text(-0.05*tlen, 0, "1000 neurons", ha="center", va="bottom", fontsize="small", rotation=90)
- cax = ax.inset_axes([0.85, -0.01, 0.15, 0.025])
- cb = plt.colorbar(im, cax=cax, orientation="horizontal")
- cb.set_ticks([0, 1.0])
- cb.set_ticklabels(["0", "1.0"], fontsize="small")
- ax.text(0.83, -0.005, "z-scored activity", fontsize="small",
- ha="right", va="top", transform=ax.transAxes)
- ax.set_ylim([-Xexs[d].shape[0]*0.05, Xexs[d].shape[0]+0.5])
- ax.set_xlim([-0.028*tlen, tlen])
- ax.axis("off")
- ax.set_title(titles[d], fontsize='medium')
- if d==0:
- il = 3 + 3*j
- transl = mtransforms.ScaledTranslation(-10 / 72, 0 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- grid1 = gridspec.GridSpecFromSubplotSpec(3, 1, subplot_spec=grid[:5, 4], wspace=0., hspace=0.3)
- for d in range(3):
- ax = fig.add_subplot(grid1[d,0])
- pos = ax.get_position().bounds
- ax.set_position([pos[0]+0.01, pos[1]+(pos[3]-pos[2]*yratio)/2+0.01,
- pos[2]*0.8, pos[2]*yratio*0.8])
- pose = ax.get_position().bounds
- ix = np.nonzero(ids==d)[0]
- if d==0:
- ix = np.hstack((ix, np.nonzero(ids==4)[0]))
- ix = np.hstack((ix, np.nonzero(ids==3)[0]))
- lns = []
- for i in ix:
- ss = evals_all[i][:1000].copy()
- alphas[i], yp = fit_powerlaw_exp(evals_all[i],
- np.arange(10, ymax))
- ss /= yp[0]
- norms[i] = yp[0] #yp[10] * 10
- ax.loglog(np.arange(1, min(len(evals_all[i])+1, 1001)),
- ss, color=colors[d], lw=0.5, alpha=0.75, ls=ls[ids[i]])
- alphas_shuff[i] = fit_powerlaw_exp(evals_shuff_all[i],
- np.arange(10, ymax))[0]
- ln = ax.plot([], [], color=colors[d], lw=1, ls=ls[ids[i]])
- lns.append(ln)
- if d==0:
- ax.legend([lns[0][0], lns[-5][0], lns[-1][0]], ["V1", "sensori-\nmotor", "PPC"],
- frameon=False, loc="upper left", bbox_to_anchor=(0.3, 1.3),
- handlelength=1.2)
- ax.set_ylim(0.003, 3)
- ax.set_xlim(1, 1000)
- ax.set_xticks([1, 10, 100, 1000])
- ax.set_xticklabels(["1", "10", "100", "1,000"])
- ax.set_yticks([0.01, 0.1 ,1])
- ax.set_yticklabels(["0.01", "0.1", "1"])
- aexp = [-0.69, -1.254]
- yc = 1
- ax.fill_between([1, 1000], [yc, yc * 1000**aexp[0]], [yc, yc * 1000**aexp[1]],
- color="k", lw=0, alpha=0.1)
- if d==0:
- il = 4
- transl_e = mtransforms.ScaledTranslation(-50 / 72, 12 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl_e)
- ax.set_ylabel("normalized variance")
- axin = ax.inset_axes([1.2, 0.5, 0.7, 0.7])
- axin.fill_between([1, 1000], [yc, yc * 1000**aexp[0]], [yc, yc * 1000**aexp[1]],
- color="k", lw=0, alpha=0.1)
- axin.text(0.27, 0.37, "symmetric", transform=axin.transAxes, fontsize="small",
- color=0.*np.ones(3), rotation=-(0.38)*90, fontstyle="italic")
- axin.text(0.08, 0.01, "non-symmetric", transform=axin.transAxes, fontsize="small",
- color=0.*np.ones(3), rotation=-(0.58)*90, fontstyle="italic")
- axin.set_xlim(1, 1000)
- axin.set_ylim(0.001, 1)
- axin.set_xscale("log")
- axin.set_yscale("log")
- axin.set_xticks([]); axin.set_yticks([])
- axin.minorticks_off()
- elif d==2:
- ax.set_xlabel("PC index")
- ax = plt.subplot(grid[:5, -2:])
- pos = ax.get_position().bounds
- ax.set_position([pos[0]+0.03, pos[1]+(pos[3]-pos[2]*yratio)/2-0.03,
- pos[2]*0.8, pos[2]*yratio*0.8])
- yh = [0, 2, 1]
- for d in range(3):
- ix = np.nonzero(ids==d)[0]
- emax = min(1000, np.array([len(evals_all[i]) for i in ix]).min())
- ev_all = np.array([evals_all[i][:emax] / norms[i] for i in ix])
- ev_mean = ev_all.mean(axis=0)
- ev_std = ev_all.std(axis=0) #/ np.sqrt(len(ix)-1)
- ax.loglog(np.arange(1, len(ev_mean)+1), ev_mean, color=colors[d], lw=1)
- ax.fill_between(np.arange(1, len(ev_mean)+1), ev_mean-ev_std, ev_mean+ev_std,
- color=colors[d], alpha=0.5, lw=0)
- ax.text(0.05, 0.1 + 0.12*yh[d], r"$\alpha$" + f" = {alphas[ix].mean():.2f}",
- transform=ax.transAxes, color=colors[d], fontsize="large")
- ax.set_ylim(0.003, 3)
- ax.set_xlim(1, 1000)
- ax.set_xticks([1, 10, 100, 1000])
- ax.set_xticklabels(["1", "10", "100", "1,000"])
- ax.set_yticks([0.01, 0.1 ,1])
- ax.set_yticklabels(["0.01", "0.1", "1"])
- ax.set_xlabel("PC index")
- ax.set_ylabel("normalized variance")
- ax.set_title("average", y=1, loc='left')
- il = 5
- transl = mtransforms.ScaledTranslation(-40 / 72, 0 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- ax = ax.inset_axes([0.7, 0.9, 0.3, 0.4])
- for d in range(3):
- ix = np.nonzero(ids==d)[0]
- if d==0:
- ix = np.hstack((ix, np.nonzero(ids==4)[0]))
- ix = np.hstack((ix, np.nonzero(ids==3)[0]))
- ash = np.stack((alphas[ix], alphas_shuff[ix]), axis=0)
- p = ttest_rel(ash[0], ash[1]).pvalue
- for i in ix:
- ax.plot(np.arange(0,2) + dy[d] * 1.5, (alphas[i], alphas_shuff[i]),
- color=colors[d], lw=1., alpha=0.75, ls=ls[ids[i]])
- star = "***" if p < 0.001 else "**" if p < 0.01 else "*" if p < 0.05 else "n.s."
- print(p)
- ax.text(dy[d]*1.5 + 0.5 + (dy[d]-1)*0., 0.9, f"{star}", ha="center",
- va="center", color=colors[d], fontsize="small" if p>=0.05 else "medium")
- ax.set_ylim([0., 1.0])
- ax.set_yticks([0., 0.5, 1.0])
- ax.set_xticks([0, 1])
- ax.set_xticklabels(["original", "shuffled"], rotation=90, ha="center", va="top")
- ax.set_ylabel("power-law\nexponent ($\\alpha$)")
- #il += 1
- #il = plot_label(ltr, il, ax, transl)
- evals_all = dat['evals_sub'].copy()
- alphas_sim = np.zeros((2, evals_all.shape[1])) * np.nan
- ax = plt.subplot(grid[-2:, 4])
- pos = ax.get_position().bounds
- ypos = pos[1]+(pos[3]-pos[2]*yratio)/2+0.03
- ax.set_position([pose[0], ypos, pos[2]*0.9, pos[2]*yratio*0.9])
- for d in range(2):
- for i in range(len(evals_all[0])):
- ss = evals_all[d, i].copy()
- alphas_sim[d, i], yp = fit_powerlaw_exp(ss, np.arange(10, ymax))
- ss /= yp[0]
- ax.loglog(np.arange(1, min(len(ss)+1, 1001)),
- ss[:1000], color=colors[d], lw=0.5, alpha=0.75)
- ax.text(0.35, 0.65 + 0.15*(1-d), r"$\alpha$" + f" = {alphas_sim[d].mean():.2f}",
- transform=ax.transAxes, color=colors[d], fontsize="medium")
- # ax.fill_between(np.arange(1, len(ev_mean)+1), ev_mean-ev_std, ev_mean+ev_std,
- # color=colors[d], alpha=0.5, lw=0)
- print(alphas_sim.mean(axis=-1))
- ax.set_title('simulated 2p data', fontsize='medium')
- ax.set_ylim(0.003, 3)
- ax.set_xlim(1, 1000)
- ax.set_xticks([1, 10, 100, 1000])
- ax.set_xticklabels(["1", "10", "100", "1,000"])
- ax.set_yticks([0.01, 0.1 ,1])
- ax.set_yticklabels(["0.01", "0.1", "1"])
- ax.set_xlabel("PC index")
- ax.set_ylabel("normalized variance")
- il += 1
- il = plot_label(ltr, il, ax, transl_e)
- transl = mtransforms.ScaledTranslation(-50 / 72, 0 / 72, fig.dpi_scale_trans)
- evals_enorms = dat['evals_enorms']
- enorms = dat['enorms']
- nnorm = len(enorms)
- n_sim = len(evals_enorms[0])
- alphas_enorms = np.zeros((nnorm, n_sim)) * np.nan
- ecolor = ['g', [0, 1, 0]]
- for d in range(nnorm):
- for i in range(len(evals_enorms[0])):
- alphas_enorms[d, i], yp = fit_powerlaw_exp(evals_enorms[d, i], np.arange(10, ymax))
- ax = plt.subplot(grid[-2:, -2])
- pos = ax.get_position().bounds
- ax.errorbar(enorms, alphas_enorms.mean(axis=1), alphas_enorms.std(axis=1)/(n_sim-1)**0.5, color=ecolor[0])
- ax1 = ax.twinx()
- ax1.set_position([pos[0] + pos[2]*0.25, ypos,
- pos[2]*0.5, pos[2]*yratio])
- ax1.plot(enorms, 0.5/(1-enorms) * 0.02, color=ecolor[1])
- ax1.spines['right'].set_visible(True)
- ax1.spines['right'].set_color(ecolor[1])
- ax1.tick_params(axis='y', colors=ecolor[1])
- ax1.set_ylabel('timescale (sec.)', color=ecolor[1], rotation=-90, va='bottom')
- ax1.set_yticks([0, 2, 4])
- ax.spines['left'].set_color(ecolor[0])
- ax.tick_params(axis='y', colors=ecolor[0])
- ax.set_xticks([0.975, 0.985, 0.998])
- ax.set_xlim([0.975, 0.998])
- ax.set_yticks([0.5, 0.6, 0.7])
- ax.set_ylabel('power-law exponent ($\\alpha$)', color=ecolor[0])
- ax.set_position([pos[0] + pos[2]*0.6, ypos,
- pos[2]*1.25, pos[2]*yratio])
- il = plot_label(ltr, il, ax, transl)
- return fig, alphas
- def suppfig_svca2(dat, dat_subsample, dat_janelia, dat_london):
- fig = plt.figure(figsize=(14, 10))
- yratio = 14 / 10
- grid = plt.GridSpec(5, 5, wspace=0.4, hspace=0.6, figure=fig,
- bottom=0.05, top=0.92, left=0.05, right=0.95)
- ix = 5
- il = 0
- dy = 0.
- shapes = dat['shapes']
- titles = ["eigenspectrum - direct", "SVCA", "SVCA2"]
- ticks = ["direct", "SVCA", "SVCA2"]
- evals_all_list = [dat["evals_all"], dat["evals_svca_all"], dat["evals_svca2_all"]]
- areas_all = dat["areas_all"]
- n_dset = len(areas_all)
- ids = np.zeros(n_dset, "int")
- alphas_all = np.zeros((n_dset, 3))
- for d in range(5):
- ids[np.array(areas_all)==areas[d]] = d
- colors = dcolors[:3].copy()
- colors = np.vstack((colors, colors[0], colors[0]))
- ls = ["-", "-", "-", "--", "-."]
- anames = ['2P cortex', '2P CA1', 'brainwide\nephys']
- grid1 = gridspec.GridSpecFromSubplotSpec(3, 3, subplot_spec=grid[:3, :3], wspace=0.2, hspace=0.4)
- for d in range(3):
- for k in range(3):
- ax = plt.subplot(grid1[d, k])
- pos = ax.get_position().bounds
- ax.set_position([pos[0] + 0.02*(1-k), pos[1], pos[3]/yratio, pos[3]])
- ix = (ids==0) + (ids==3) + (ids==4) if d==0 else ids==d
- ix = np.nonzero(ix)[0]
- for i in ix:
- evals = evals_all_list[k][i].copy()
- alphas_all[i, k], yp = fit_powerlaw_exp(evals, np.arange(10, 500))
- evals /= yp[0]
- ax.loglog(np.arange(1, len(evals)+1), evals, color=colors[ids[i]],
- lw=1, alpha=0.25 if d!=3 else 0.75, ls=ls[ids[i]], zorder=-30*(ids[i]==1) + 20*(ids[i]!=1))
- ax.minorticks_on()
- ax.set_ylim(0.001, 3)
- ax.set_xlim(1, 3000)
- ax.set_yticks([0.01, 0.1, 1])
- ax.set_yticklabels(["0.01", "0.1", "1"], fontsize='small')
- ax.set_xticks([1, 10, 100, 1000])
- ax.set_xticklabels(['1', '10 ', '100 ', ' 1,000'], fontsize='small')
- ax.text(0.3, 0.85, "$\\alpha = $" + f"{alphas_all[ix, k].mean():.2f}",
- color=colors[d], transform=ax.transAxes, fontweight="bold")
- ax.xaxis.set_minor_locator(matplotlib.ticker.LogLocator(base=10, subs=np.arange(2,10), numticks=10))
- if k==0:
- ax.text(0.05, 0.05, anames[d], color=colors[d],
- transform=ax.transAxes, fontweight='bold', fontstyle='italic')
- transl = mtransforms.ScaledTranslation(-40 / 72, 5 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- if d==0:
- if k==0:
- ax.set_xlabel("PC index")
- ax.set_ylabel("normalized variance")
- #ax.set_title("neural recordings", loc="left", x=-0.1, fontweight="bold")
- ax.set_title(titles[k], y=1.1, fontstyle='italic')
- ax = plt.subplot(grid[:3, 3])
- pos = ax.get_position().bounds
- ax.set_position([pos[0]- 0.03, pos[1]+0.25*pos[3], pos[2], pos[3]*0.8])
- xp = np.arange(3)*np.ones((len(evals_all_list[0]),1))
- xp += np.random.randn(*xp.shape)*0.05
- cols = np.tile(colors[ids][:,np.newaxis], (1,3)).reshape(-1,3)
- ax.scatter(xp.flatten(), alphas_all.flatten(), color=cols, s=10)
- for d in range(3):
- ix = (ids==0) + (ids==4) + (ids==5) if d==0 else ids==d
- ax.scatter(np.arange(3), alphas_all[ix].mean(axis=0), color=colors[d],
- s=400, marker="_")
- ax.set_ylabel("power-law exponent ($\\alpha$)")
- ax.set_xticks(np.arange(3))
- ax.set_xticklabels(["direct", "SVCA", "SVCA2"], rotation=45, ha='right')
- transl = mtransforms.ScaledTranslation(-50 / 72, -5 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- ax.set_ylim([0.2, 1.4])
- ax = plt.subplot(grid[:3, 4])
- cols = [colors[0], colors[0], [0.8, 0, 0], [0.5, 0, 0.5], [1, 0, 1]]
- pos = ax.get_position().bounds
- ax.set_position([pos[0]-0.01, pos[1]+0.25*pos[3], pos[2]*1.25, pos[3]*0.8])
- ix = (ids==0) + (ids==4) + (ids==5)
- for j in range(5):
- if j<2:
- alphas = alphas_all[ix, (1-j)+1]
- elif j==2:
- alphas = dat_subsample['alphas_all']
- elif j==3:
- alphas = dat_janelia['alphas_all']
- elif j==4:
- alphas = dat_london['alphas_all']
- xp = j*np.ones(len(alphas))
- xp += np.random.randn(*xp.shape)*(0.05 + 0.05*(j==4))
- ax.scatter(xp, alphas, color=cols[j], s=10)
- ax.scatter(j, alphas.mean(), color=cols[j], s=500, marker="_")
- ax.set_ylabel("power-law exponent ($\\alpha$)")
- ax.set_ylim([0.2, 1.4])
- ax.set_xticks(np.arange(5))
- ax.set_xticklabels(["SVCA2, GCaMP8s 22Hz", "SVCA, GCaMP8s 22Hz",
- "SVCA, GCaMP8s 3Hz", "SVCA, GCaMP6s 3Hz\n(Janelia)",
- "SVCA, GCaMP6s 3Hz\n(London)"], rotation=45, ha='right')
- for idx, lbl in enumerate(ax.get_xticklabels()):
- lbl.set_color(cols[idx])
- transl = mtransforms.ScaledTranslation(-50 / 72, -5 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- shapes = dat['shapes']
- areas_all = dat['areas_all']
- n_dset = len(shapes)
- ids = np.zeros(n_dset, "int")
- for d in range(5):
- ids[np.array(areas_all)==areas[d]] = d
- anames = ['2P cortex', '2P CA1', 'brainwide ephys']
- scolors = ['g', 'y', 'b']
- bsize = 4000
- ikeeps = np.arange(50, bsize+1, 200)
- ikeeps[-1] = 4000
- ineurs = np.arange(0.01, 1.05, 0.05)
- ineurs[-1] = 1.0
- nneurons = ineurs * np.array(shapes)[:,:1]
- ntimes = ikeeps / bsize * np.array(shapes)[:,1:] / (22*60)
- xticks = [[[0, 2500, 5000], [0, 50, 100]],
- [[0, 2500, 5000], [0, 50, 100]],
- [[0, 1000, 2000], [0, 15, 30]]]
- fstr = ['', 'svca_', 'svca2_']
- grid1 = gridspec.GridSpecFromSubplotSpec(1, 6, subplot_spec=grid[-2:, :],
- wspace=0.4, hspace=0.4)
- axs = [[plt.subplot(grid1[-2:, 2*d]) for d in range(3)],
- [plt.subplot(grid1[-2:, 2*d+1]) for d in range(3)]]
- for k in range(3):
- evals_all = [dat[f'evals_{fstr[k]}all_times'].copy(), dat[f'evals_{fstr[k]}all_neurons'].copy()]
- for j in range(2):
- nvar = len(evals_all[j][0])
- alphas_all = np.zeros((n_dset, nvar))*np.nan
- for i in range(len(evals_all[j])):
- for n in range(len(evals_all[j][i])):
- evals = evals_all[j][i][n].copy()
- ymax = min(len(evals)//2, 500) #500 if len(evals) > 500 else min(len(evals), 100)
- if ymax > 15:
- alphas_all[i,n], yp = fit_powerlaw_exp(evals, np.arange(10, ymax))
- for d in range(3):
- ax = axs[j][d]
- ix = (ids==0) + (ids==4) + (ids==5) if d==0 else ids==d
- nn = nneurons[ix].mean(axis=0) if j==1 else ntimes[ix].mean(axis=0)
- ax.errorbar(nn, np.nanmean(alphas_all[ix], axis=0),
- np.nanstd(alphas_all[ix], axis=0) / ((~np.isnan(alphas_all[ix])).sum(axis=0)-1)**0.5,
- color=scolors[k], lw=1)
- ax.set_ylim([0.2, 1.5])
- ax.tick_params(axis='both', labelsize='small')
- ax.set_xticks(xticks[d][1-j])
- ax.set_yticks([0.5, 1, 1.5])
- ax.set_xlabel('# of neurons' if j==1 else 'duration (min.)')
- if k==0 and j==0:
- ax.set_ylabel('power-law exponent ($\\alpha$)')
- transl = mtransforms.ScaledTranslation(-55 / 72, -5 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- ax.set_title(anames[d], loc='left', color=colors[d],
- fontstyle='italic', fontweight='bold')
- if k==0:
- pos = ax.get_position().bounds
- ax.set_position([pos[0]+0.015*(j==0)*(k==0), pos[1], pos[2], pos[3]*0.9])
- if j==0 and d==0:
- ax.text(1, 0.9-0.1*k, ['direct', 'SVCA', 'SVCA2'][k], color=scolors[k],
- ha='right', transform=ax.transAxes)
- return fig
- def suppfig_ephys(dat, dat_areas, dat_tbins):
- fig = plt.figure(figsize=(8, 9))
- yratio = 8 / 9
- grid = plt.GridSpec(7, 4, wspace=0.45, hspace=0.6, figure=fig,
- bottom=0.07, top=0.98, left=0.1, right=0.95)
- ix = 5
- il = 0
- dy = 0.
- shapes = dat['shapes']
- areas_all = dat["areas_all"]
- n_dset = len(areas_all)
- ids = np.zeros(n_dset, "int")
- alphas_all = np.zeros((n_dset, 3))
- for d in range(5):
- ids[np.array(areas_all)==areas[d]] = d
- ax = plt.subplot(grid[:3, :])
- pos = ax.get_position().bounds
- ax.set_position([pos[0]-0.05, pos[1]-0.2*pos[3], pos[2]+0.065, pos[3]*1.08])
- tmin = 1120#42000//7 - 0
- tlen = 22*60*3//7
- Xex = zscore(dat['Xexs'][2][:, tmin:tmin+tlen], axis=1)
- im = ax.imshow(Xex, aspect="auto", cmap="gray_r", vmin=0., vmax=1.25)
- ax.plot([0, 22*30/7], -Xex.shape[0]*0.03*np.ones(2), color="k", lw=1.5)
- ax.plot(-0.015*tlen*np.ones(2), [0, 50], color="k", lw=1.5)
- ax.text((22*30/2)/7, -0.07*Xex.shape[0], "30 sec.", ha="center", va="center", fontsize="small")
- ax.text(-0.03*tlen, 0, "500 neurons", ha="center", va="bottom", fontsize="small", rotation=90)
- cax = ax.inset_axes([0.85, -0.01, 0.15, 0.025])
- cb = plt.colorbar(im, cax=cax, orientation="horizontal")
- cb.set_ticks([0, 1.0])
- cb.set_ticklabels(["0", "1.0"], fontsize="small")
- ax.text(0.83, -0.005, "z-scored activity", fontsize="small",
- ha="right", va="top", transform=ax.transAxes)
- ax.set_ylim([-Xex.shape[0]*0.05, Xex.shape[0]+0.5])
- ax.set_xlim([-0.028*tlen, tlen])
- ax.axis("off")
- ax.set_title('Rastermap of brainwide ephys activity')
- transl = mtransforms.ScaledTranslation(-20 / 72, 0 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- colors = dcolors[:3].copy()
- acolors = ['g', 'm', [0.25, 1, 0.25], 'r', [0.5, 0, 0]]
- ineurs = np.arange(0.01, 1.05, 0.05)
- ineurs[-1] = 1.0
- nneurons = ineurs * np.array(shapes)[:,:1]
- evals_all = dat[f'evals_svca2_all_neurons'].copy()
- nvar = len(evals_all[0])
- alphas_all = np.zeros((n_dset, nvar))*np.nan
- for i in range(len(evals_all)):
- for n in range(len(evals_all[i])):
- evals = evals_all[i][n].copy()
- ymax = min(len(evals)//2, 500) #500 if len(evals) > 500 else min(len(evals), 100)
- if ymax > 15:
- alphas_all[i,n], yp = fit_powerlaw_exp(evals, np.arange(10, ymax))
- nh = [4, 2, 3, 1, 0]
- area_names = dat_areas['area_groups'].keys()
- evals_all = dat_areas['evals_areas'].copy()
- nareas = len(area_names)
- alphas_areas = np.zeros((nareas, 3)) * np.nan
- nn_areas = np.stack([dat_areas['nneurons'][a] for a in area_names], axis=0)
- for i in range(3):
- ax = plt.subplot(grid[3:5, i+1])
- pos = ax.get_position().bounds
- ax.set_position([pos[0] + (2-i)*0.03 + 0.01, pos[1]-0.02, pos[2]*0.95, pos[2]*yratio*0.95])
- for n, area in enumerate(area_names):
- evals = evals_all[area][i].copy()
- if len(evals) > 0:
- ymax = min(len(evals)//2, 500) #500 if len(evals) > 500 else min(len(evals), 100)
- alphas_areas[n, i], yp = fit_powerlaw_exp(evals, np.arange(10, ymax))
- evals /= yp[0]
- ax.loglog(np.arange(1, len(evals)+1), evals, color=acolors[n], lw=1)
- nstr = '# of neurons ' if nh[n]==0 else ''
- nhn = nh[n] if i<2 else nh[n] - 1*(nh[n]>1)
- ax.text(1, 0.95-0.12*nhn, f'{nstr}= {int(nn_areas[n, i]):,d}', color=acolors[n],
- ha='right', transform=ax.transAxes, fontsize='small')
- ax.set_ylim(0.003, 3)
- ax.set_xlim(1, 1000)
- ax.set_yticks([0.01, 0.1, 1])
- ax.set_yticklabels(["0.01", "0.1", "1"], fontsize='small')
- ax.set_xticks([1, 10, 100, 1000])
- ax.set_xticklabels(['1', '10', '100', '1,000'], fontsize='small')
- ax.set_title(f'mouse {i+1}', fontsize='medium', y=1.1)
- ax.set_xlabel("PC index")
- if i==0:
- ax.set_ylabel("normalized variance")
- transl = mtransforms.ScaledTranslation(-50 / 72, 10 / 72, fig.dpi_scale_trans)
- il = 2
- il = plot_label(ltr, il, ax, transl)
- il -= 2
- d = 2
- ax = plt.subplot(grid[3:5, 0])
- pos = ax.get_position().bounds
- ax.set_position([pos[0], pos[1]-0.02, pos[2], pos[2]*yratio*1.1])
- ix = ids==d
- nn = nneurons[ix].mean(axis=0)
- ax.errorbar(nn, np.nanmean(alphas_all[ix], axis=0),
- np.nanstd(alphas_all[ix], axis=0) / ((~np.isnan(alphas_all[ix])).sum(axis=0)-1)**0.5,
- color=colors[d], lw=1)
- for n, area in enumerate(area_names):
- ax.scatter(nn_areas[n], alphas_areas[n], color=acolors[n],
- s=30, marker='x', zorder=30, alpha=0.9)
- ax.text(1.25, 1-0.12*nh[n], area, color=acolors[n], ha='right',
- transform=ax.transAxes, fontsize='small', va='bottom')
- ax.text(1.25, 0.02, u'\u2014 random\nsubsets', color=dcolors[d], ha='right',
- transform=ax.transAxes, fontweight='bold')
- ax.set_ylim([0.4, 1.8])
- ax.set_xscale('log')
- ax.set_xticks([100, 1000])
- ax.set_xticklabels(['100', '1,000'])
- ax.set_xlabel('# of neurons')
- ax.set_ylabel('power-law exponent ($\\alpha$)')
- transl = mtransforms.ScaledTranslation(-55 / 72, 0 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- il += 1
- tbins = dat_tbins['tbins']
- acolors = plt.get_cmap('YlOrBr_r')(np.linspace(0, 0.6, len(tbins)))
- evals_all = dat_tbins['evals_tbins']
- alphas_tbins = np.zeros((len(tbins), 3)) * np.nan
- for i in range(3):
- ax = plt.subplot(grid[5:7, i+1])
- pos = ax.get_position().bounds
- ax.set_position([pos[0] + (2-i)*0.03 + 0.01, pos[1], pos[2]*0.95, pos[2]*yratio*0.95])
- for n, tbin in enumerate(tbins):
- evals = evals_all[i][n].copy()
- if len(evals) > 0:
- ymax = min(len(evals)//2, 500) #500 if len(evals) > 500 else min(len(evals), 100)
- alphas_tbins[n, i], yp = fit_powerlaw_exp(evals, np.arange(10, ymax))
- evals /= yp[0]
- ax.loglog(np.arange(1, len(evals)+1), evals, color=acolors[n], lw=1)
- ax.set_ylim(0.003, 3)
- ax.set_xlim(1, 1000)
- ax.set_yticks([0.01, 0.1, 1])
- ax.set_yticklabels(["0.01", "0.1", "1"], fontsize='small')
- ax.set_xticks([1, 10, 100, 1000])
- ax.set_xticklabels(['1', '10', '100', '1,000'], fontsize='small')
- ax.set_title(f'mouse {i+1}', fontsize='medium')
- ax.set_xlabel("PC index")
- if i==0:
- ax.set_ylabel("normalized variance")
- transl = mtransforms.ScaledTranslation(-50 / 72, 10 / 72, fig.dpi_scale_trans)
- il += 1
- il = plot_label(ltr, il, ax, transl)
- il -= 2
- ax = plt.subplot(grid[5:7, 0])
- pos = ax.get_position().bounds
- ax.set_position([pos[0], pos[1], pos[2], pos[2]*yratio*1.1])
- ax.errorbar(tbins*1000, alphas_tbins.mean(axis=-1),
- alphas_tbins.std(axis=-1)/(2**0.5), color='k')
- for n, tbin in enumerate(tbins):
- ax.scatter(tbin*1000, alphas_tbins[n].mean(), s=30,
- color=acolors[n], zorder=30)
- ax.set_xscale('log')
- ax.set_xlabel('time bin (ms)')
- ax.set_ylabel('power-law exponent ($\\alpha$)')
- ax.set_xticks([10, 100])
- ax.set_xticklabels(['10', '100'])
- ax.set_ylim([0.65, 0.85])
- ax.set_yticks([0.7, 0.8])
- transl = mtransforms.ScaledTranslation(-58 / 72, 0 / 72, fig.dpi_scale_trans)
- il = plot_label(ltr, il, ax, transl)
- return fig
- def suppfig_running(dat):
- colors = [[0.7, 0, 0], [1, 0.5, 0.5]]
- fig = plt.figure(figsize=(14, 3), dpi=150)
- yratio = 14 / 3
- grid = plt.GridSpec(1, 5, wspace=0.4, hspace=0.2, figure=fig,
- bottom=0.18, top=0.9, left=0.05, right=0.95)
- transl = mtransforms.ScaledTranslation(-45 / 72, 5 / 72, fig.dpi_scale_trans)
- il = 0
- evals_run_all = dat['evals_run_all'].copy()
- nrun = evals_run_all.shape[1]
- alphas_all = np.zeros((2, nrun))
- ax = plt.subplot(grid[0, 0])
- pos = ax.get_position().bounds
- ax.set_position([pos[0], pos[1], pos[2], pos[2]*yratio])
- rstr = ['running', 'not running']
- for j in range(2):
- for i in range(nrun):
- alphas_all[j, i], yp = fit_powerlaw_exp(evals_run_all[j, i], np.arange(10, 500))
- evals_run_all[j, i] /= yp[0]
- evals = np.nanmean(evals_run_all[j], axis=0)
- evals_std = np.nanstd(evals_run_all[j], axis=0) / (nrun-1)**0.5
- ax.loglog(np.arange(1, len(evals)+1), evals, lw=1, color=colors[j])
- ax.fill_between(np.arange(1, len(evals)+1), evals - evals_std, evals + evals_std,
- edgecolor='none', facecolor=colors[j], alpha=0.25)
- alpha = fit_powerlaw_exp(evals, np.arange(10, 500))[0]
- ax.text(1, 1-0.1*j, f'{rstr[j]}, $\\alpha$={alpha:.2f}', ha='right',
- transform=ax.transAxes, color=colors[j])
- ax.set_yscale('log')
- ax.set_xscale('log')
- ax.set_ylim(0.003, 0.003*700)
- ax.set_xlim(1, 700)
- ax.set_yticks([0.01, 0.1, 1])
- ax.set_yticklabels(["0.01", "0.1", "1"], fontsize='small')
- ax.set_xticks([1, 10, 100])
- ax.set_xticklabels(['1', '10', '100'], fontsize='small')
- ax.set_xlabel("PC index")
- ax.set_ylabel("normalized variance")
- il = plot_label(ltr, il, ax, transl)
- ax = plt.subplot(grid[0, 1])
- pos = ax.get_position().bounds
- ax.set_position([pos[0] + pos[2]*0.1, pos[1], pos[2]*0.5, pos[2]*yratio])
- areas = dat['areas']
- ls = {'V1': '-', 'sensorimotor': '--', 'PPC': '-.'}
- for i in range(nrun):
- ax.plot(np.arange(0,2), alphas_all[:,i],
- color='k', lw=1., alpha=0.75, ls=ls[areas[i]])
- p = ttest_rel(alphas_all[0], alphas_all[1]).pvalue
- star = "***" if p < 0.001 else "**" if p < 0.01 else "*" if p < 0.05 else "n.s."
- print(p)
- ax.text(0.5, 1., f"{star}", ha="center",
- va="center", color='k', fontsize="small" if p>=0.05 else "medium")
- ax.set_ylim([0., 1.0])
- ax.set_yticks([0., 0.5, 1.0])
- ax.set_xticks([0, 1])
- ax.set_xticklabels(["running", "not\nrunning"], rotation=0, ha="center", va="top")
- ax.set_ylabel("power-law exponent ($\\alpha$)")
- lns = []
- area_names = ['V1', 'sensorimotor', 'PPC']
- for area in area_names:
- ln = ax.plot([], [], color='k', lw=1, ls=ls[area])
- lns.append(ln[0])
- ax.legend(lns, area_names,
- frameon=False, loc="lower left", #bbox_to_anchor=(0.2, 1.1),
- handlelength=1.2)
- il = plot_label(ltr, il, ax, transl)
- iex = 2
- for j in range(2):
- ax = plt.subplot(grid[0, 2+j])
- pos = ax.get_position().bounds
- ax.set_position([pos[0], pos[1], pos[2], pos[2]*yratio])
- e = dat['dmd_evals_all'][j, iex].copy()
- ix = np.abs(e)>.25
- iang = np.angle(e[ix]) / (2*np.pi)
- iabs = -np.log10(np.abs(e[ix]))
- irot = iang/iabs
- ixx = e[ix].imag>=0
- mu = irot[ixx].mean()
- sd = irot[ixx].std()
- m = np.percentile(irot[ixx], [5, 25, 75, 95])
- med = np.median(irot[ixx])
- print(m, med)
- ax.scatter(e.real, e.imag, s=10, color=colors[j])
- ax.set_ylim([-1, 1])
- ax.set_xlim([0.25, 1])
- ax.set_xlabel('real part')
- ax.set_ylabel('imaginary part')
- ax.text(0.5, 0.8, rstr[j], color=colors[j], ha='center', transform=ax.transAxes)
- if j==0:
- il = plot_label(ltr, il, ax, transl)
- ax.set_title('Eigenvalues of DMD matrix (dt=0.23s), example mouse', loc='left', y=1.02, fontsize='medium')
- ax = plt.subplot(grid[0, -1])
- pos = ax.get_position().bounds
- ax.set_position([pos[0], pos[1], pos[2], pos[2]*yratio])
- igood = np.nonzero(~np.isnan(dat['dmd_evals_all'][0,:,0]))[0]
- for i, iex in enumerate(igood):
- for j in range(2):
- e = dat['dmd_evals_all'][j, iex].copy()
- ix = np.abs(e)>.25
- iang = np.angle(e[ix]) / (2*np.pi)
- iabs = -np.log10(np.abs(e[ix]))
- irot = iang/iabs
- ixx = e[ix].imag>=0
- mu = irot[ixx].mean()
- sd = irot[ixx].std()
- m = np.percentile(irot[ixx], [5, 25, 50, 75, 95])
- med = np.median(irot[ixx])
- dy = 0.3
- yy = -(4*i - j)
- print(ixx.sum())
- ax.plot([m[0], m[-1]], -(4*i - j) * np.ones(2), lw=1, color=np.minimum(1, np.array(colors[j])+0.25))
- ax.plot([m[1], m[-2]], -(4*i - j) * np.ones(2), lw=4, color=np.minimum(1, np.array(colors[j])+0.25))
- ax.plot([m[2], m[2]], [yy-dy, yy+dy], lw=1, color=colors[j])
- if j==0:
- ax.text(-0.1, -4*i, f'mouse {len(igood) - i}', rotation=90, ha='center', va='center', fontsize='small')
- ax.set_xlim([-0.1, 1])
- ax.set_ylim([3, -9])
- ax.spines['left'].set_visible(False)
- ax.set_yticks([])
- ax.set_xlabel('rotations per 10-fold attenuation')
- il = plot_label(ltr, il, ax, transl)
- return fig
- # def suppfig_svca(dat):
- # fig = plt.figure(figsize=(14, 14), dpi=150)
- # yratio = 14 / 14
- # grid = plt.GridSpec(5, 6, wspace=0.4, hspace=0.2, figure=fig,
- # bottom=0.05, top=1, left=0.08, right=0.95)
- # ix = 5
- # transl = mtransforms.ScaledTranslation(-45 / 72, ix / 72, fig.dpi_scale_trans)
- # il = 0
- # dy = 0.
- # titles = ["eigenspectrum - direct", "SVCA", "SVCA2"]
- # ticks = ["direct", "SVCA", "SVCA2"]
- # evals_all_list = [dat["evals_all"], dat["evals_svca_all"], dat["evals_svca2_all"]]
- # areas_all = dat["areas_all"]
- # ids = np.zeros(len(areas_all), "int")
- # alphas_all = np.zeros((len(evals_all_list[0]), 3))
- # for d in range(5):
- # ids[np.array(areas_all)==areas[d]] = d
- # colors = dcolors[:3].copy()
- # colors.append(colors[0])
- # colors.append(colors[0])
- # ls = ["-", "-", "-", "--", "-."]
- # for d in range(5):
- # ax = plt.subplot(grid[-1, d])
- # pos = ax.get_position().bounds
- # ax.set_position([pos[0], pos[1], pos[2], pos[2]*yratio])
- # for i in range(len(evals_all_list[d])):
- # evals = evals_all_list[d][i].copy()
- # alphas_all[i, d], yp = fit_powerlaw_exp(evals, np.arange(10, 500))
- # evals /= yp[0]
- # ax.loglog(np.arange(1, len(evals)+1), evals, color=dcolors[ids[i]],
- # lw=1, alpha=1, ls=ls[d])
- # ax.set_ylim(0.001, 3)
- # ax.set_xlim(1, 3000)
- # ax.set_yticks([0.01, 0.1, 1])
- # ax.set_yticklabels(["0.01", "0.1", "1"])
- # ax.set_xticks([1, 10, 100, 1000])
- # ax.set_xticklabels(["1", "10", "100", "1,000"])
- # for k in range(3):
- # ax.text(0.05, 0.32 - 0.12*k, "$\\alpha = $" + f"{alphas_all[ids==k, d].mean():.2f}",
- # color=dcolors[k], transform=ax.transAxes, fontweight="bold")
- # ax.set_xlabel("PC index")
- # if d==0:
- # ax.set_ylabel("normalized variance")
- # il = plot_label(ltr, il, ax, transl)
- # ax.set_title("neural recordings", loc="left", x=-0.1, fontweight="bold")
- # ax = plt.subplot(grid[-1, 3])
- # pos = ax.get_position().bounds
- # ax.set_position([pos[0] + 0.025, pos[1]+dy-0.02, pos[2], pos[2]*yratio + 0.04])
- # xp = np.arange(3)*np.ones((len(evals_all_list[0]),1))
- # xp += np.random.randn(*xp.shape)*0.05
- # cols = np.tile(dcolors[ids][:,np.newaxis], (1,3)).reshape(-1,3)
- # ax.scatter(xp.flatten(), alphas_all.flatten(), color=cols, s=10)
- # for k in range(3):
- # ax.scatter(np.arange(3), alphas_all[ids==k].mean(axis=0), color=dcolors[k],
- # s=400, marker="_")
- # ax.set_ylabel("power-law exponent ($\\alpha$)")
- # ax.set_xticks(np.arange(3))
- # ax.set_xticklabels(["direct", "SVCA", "SVCA2"], rotation=0)
- # il = plot_label(ltr, il, ax, transl)
- # return fig
- # def suppfig_svca2_sizes(dat_sim, alphas=None):
- # evals_svca2_all = dat_sim["evals_svca2_all"]
- # nonsyms = dat_sim["nonsyms"]
- # nneurons = dat_sim["nneurons"]
- # ntimes = dat_sim["ntimes"].astype('float32') / 23 # convert to seconds
- # noise_levels = dat_sim["noise_levels"]
- # n_sim = len(evals_svca2_all)
- # if alphas is None:
- # alphas = np.zeros((evals_svca2_all.shape[:-1]))
- # for i in range(n_sim):
- # for ni, nonsym in enumerate(nonsyms):
- # for nl, noise_level in enumerate(noise_levels):
- # for ii, nneur in enumerate(nneurons):
- # for jj, ntime in enumerate(ntimes):
- # evals = evals_svca2_all[i,ni,nl,ii,jj].copy()
- # ymax = min((~np.isnan(evals)).sum(), int(nneur*0.4), int(ntime*0.4))
- # evals = evals_svca2_all[i,ni,nl,ii,jj].copy()
- # yrange = np.arange(10, min(500, (~np.isnan(evals)).sum()-1))
- # alpha = fit_powerlaw_exp(evals, yrange)[0]
- # alphas[i,ni,nl,ii,jj] = alpha
- # fig = plt.figure(figsize=(14, 9), dpi=150)
- # yratio = 14 / 9
- # grid = plt.GridSpec(4, 7, wspace=0.6, hspace=0.7, figure=fig,
- # bottom=0.05, top=0.95, left=0.06, right=0.98)
- # grid1 = gridspec.GridSpecFromSubplotSpec(4, 6, subplot_spec=grid[:, 2:], wspace=0.3, hspace=0.7)
- # il = 0
- # n_sim = len(evals_svca2_all)
- # for ni, nonsym in enumerate(nonsyms):
- # for j in range(2):
- # if j==0:
- # lcolors = plt.get_cmap("Oranges")(np.linspace(0.5, 1, len(nneurons)//2))
- # else:
- # lcolors = plt.get_cmap("Greens")(np.linspace(0.5, 1, len(ntimes)//2))
- # ax = plt.subplot(grid[ni, j])
- # pos = ax.get_position().bounds
- # ax.set_position([pos[0], pos[1], pos[2], pos[2]*yratio])
- # if ni==0:
- # transl = mtransforms.ScaledTranslation(-40 / 72, -4 / 72, fig.dpi_scale_trans)
- # else:
- # transl = mtransforms.ScaledTranslation(-40 / 72, -10 / 72, fig.dpi_scale_trans)
- # il = plot_label(ltr, il, ax, transl)
- # if j==0:
- # if ni==0:
- # x0, y0 = -0.65, 1.3
- # else:
- # x0, y0 = -0.65, 1.2
- # ax.text(x0, y0, ["symmetric", "1/3 non-symmetric", "2/3 non-symmetric", "non-symmetric"][ni] + " connectivity",
- # transform=ax.transAxes, fontsize="medium", fontweight="bold", fontstyle='italic')
- # if ni == 0:
- # ax.text(0.7, 1.05, '# neurons = ' if j==0 else 'time =',
- # color='k', transform=ax.transAxes, ha='right')
- # for ii, nn in enumerate(nneurons) if j ==0 else enumerate(ntimes):
- # if ii%2==1:
- # continue
- # evals = evals_svca2_all[0,ni,0,ii,-1].copy() if j==0 else evals_svca2_all[0,ni,0,-1,ii].copy()
- # ymax = min((~np.isnan(evals)).sum(), int(nn*0.4))
- # evals = evals[:ymax]
- # alpha = plot_spectrum(ax, evals, color=lcolors[ii//2], lw=1, plot_fit=False)
- # if ii%4==0 and ni==0:
- # if j==0:
- # ax.text(0.75, 0.65 + 0.1*ii/4, f"{nn:.0f}",
- # color=lcolors[ii//2], transform=ax.transAxes, fontsize="small")
- # else:
- # ax.text(0.75, 1.05 - 0.1*ii/4, f"{nn/60:.1f} min",
- # color=lcolors[ii//2], transform=ax.transAxes, fontsize="small")
- # if ni==0:
- # if j==0:
- # ax.set_xlabel("PC index")
- # ax.set_ylabel("normalized\nvariance")
- # else:
- # ax.set_xlabel("PC index")
- # for j in range(len(noise_levels)):
- # ax = plt.subplot(grid1[ni, j])
- # pos = ax.get_position().bounds
- # ax.set_position([pos[0]+0.008*(len(noise_levels) - j), pos[1], pos[2], pos[2]*yratio])
- # im = ax.imshow(alphas.mean(axis=0)[ni, j].T, vmin=0.5, vmax=2, cmap='viridis',
- # aspect='auto')
- # #ax.invert_yaxis()
- # xticks = np.array(1 * 2**np.arange(0, 9, 2))
- # f = interp1d(ntimes, np.arange(len(ntimes)))
- # ax.set_xticks(f(xticks * 60))
- # yticks = np.array(200 * 2**np.arange(0, 6, 1))
- # yticks = np.hstack((yticks, 10000))
- # f = interp1d(nneurons, np.arange(len(nneurons)))
- # ax.set_yticks(f(yticks))
- # ax.tick_params(labelsize='small')
- # #ax.axis('square')
- # ax.set_title(['low', 'medium', 'high'][j%3], fontsize='small')
- # if j==1:
- # ax.text(0.5, 1.2, 'Gaussian noise + smoothing', fontsize='medium',
- # ha='center', transform=ax.transAxes, fontstyle='italic')
- # elif j==4:
- # ax.text(0.5, 1.2, 'Poisson noise', fontsize='medium',
- # ha='center', transform=ax.transAxes, fontstyle='italic')
- # if ni==0:
- # if j==1:
- # axin = ax.inset_axes([0.05, -0.25, 0.8, 0.1]);
- # cb = plt.colorbar(im, cax=axin, orientation='horizontal')
- # cb.ax.tick_params(labelsize='small')
- # axin.text(1.05, 0.5, 'power-law exponent ($\\alpha$)', fontsize='small',
- # ha='left', transform=axin.transAxes, va='center')
- # cb.ax.set_xticks([0.5, 1, 1.5, 2])
- # #cb.ax.yaxis.label.set_size('medium')
- # elif j==0:
- # ax.set_ylabel('# of neurons')
- # ax.set_xlabel('time (min)')
- # if j==0:
- # transl = mtransforms.ScaledTranslation(-50 / 72, -5 / 72, fig.dpi_scale_trans)
- # il = plot_label(ltr, il, ax, transl)
- # ax.set_yticklabels(yticks)
- # ax.set_xticklabels(xticks, rotation=30, ha='right', va='top')
- # else:
- # ax.set_yticklabels([])
- # ax.set_xticklabels([])
- # return fig
fig2.py at commit 2c15edf, under GPL-3.0 · at the source
Overview
Abstract
Intrinsically generated, brainwide neural activity displays macroscopic coordination among large populations of neurons that persists beyond the biophysical timescales of individual neurons1–3. It is not well understood how these macroscopic behaviours arise from microscopic, short-lived interactions between pairs of neurons. Here we show that the eigenvalue spectrum and dynamical properties of large-scale neural recordings in mice are similar to those produced by linear dynamics governed by a random symmetric matrix that is critically normalized. An exception was population activity in hippocampal area CA1, which resembled an efficient, uncorrelated neural code that may be optimized for information storage capacity. High-dimensional, global activity modes emerged in critically normalized artificial networks and persisted under sparse, clustered or spatial connectivity. These dynamics were useful for solving time-dependent tasks such as a zero-shot working memory task.
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 13 matches between paragraphs and lines of code.
mouseland/critical_init
2c15edf4e770165fc3962dcc3920c8bcaf555bed, 4 June 2026Availability: 1 check, the latest on 28 September 2026: the link answers
- 28 September 2026: the link answers
22 files
- demo.ipynb, Jupyter, 98 lines
- fig1/
fig1.ipynb , Jupyter, 115 lines - fig1/
fig1.py , Python, 602 lines - fig2/
analysis.py , Python, 534 lines, 1 match - fig2/
fig2.ipynb , Jupyter, 142 lines - fig2/
fig2.py , Python, 973 lines, 2 matches - fig3/
fig3.ipynb , Jupyter, 174 lines - fig3/
fig3.py , Python, 382 lines - fig3/
other_datasets.ipynb , Jupyter, 74 lines - fig3/
other_datasets.py , Python, 659 lines, 1 match - fig4/
analysis.py , Python, 184 lines - fig4/
fig4.ipynb , Jupyter, 103 lines - fig4/
fig4.py , Python, 364 lines - fig4/
sparse_clustered_local_s , Python, 298 lines, 2 matchesims.py - fig5/
fig5.ipynb , Jupyter, 163 lines, 2 matches - fig5/
fig5.py , Python, 443 lines - fig_utils.py, Python, 75 lines, 1 match
- lyapun.py, Python, 145 lines
- powerlaw.py, Python, 187 lines
- simulations.py, Python, 382 lines, 1 match
- LICENSE, License, 674 lines
- README.md, Text, 51 lines
Zenodo 19322086
Availability: 1 check, the latest on 28 September 2026: the link answers (HTTP 200)
- 28 September 2026: the link answers (HTTP 200)
22 files
- demo.ipynb, Jupyter, 98 lines
- fig1/
fig1.ipynb , Jupyter, 115 lines - fig1/
fig1.py , Python, 602 lines - fig2/
analysis.py , Python, 534 lines - fig2/
fig2.ipynb , Jupyter, 142 lines - fig2/
fig2.py , Python, 969 lines, 2 matches - fig3/
fig3.ipynb , Jupyter, 174 lines - fig3/
fig3.py , Python, 382 lines - fig3/
other_datasets.ipynb , Jupyter, 77 lines - fig3/
other_datasets.py , Python, 656 lines, 1 match - fig4/
analysis.py , Python, 184 lines - fig4/
fig4.ipynb , Jupyter, 103 lines - fig4/
fig4.py , Python, 364 lines - fig4/
sparse_clustered_local_s , Python, 298 linesims.py - fig5/
fig5.ipynb , Jupyter, 163 lines - fig5/
fig5.py , Python, 443 lines - fig_utils.py, Python, 75 lines
- lyapun.py, Python, 145 lines
- powerlaw.py, Python, 187 lines
- simulations.py, Python, 382 lines
- LICENSE, License, 674 lines
- README.md, Text, 49 lines
Code availability
Code to reproduce all the analyses and figures is available at GitHub (https://
Reproduced under the paper's license (CC BY), from the paper cited above.
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:
- 2 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
- 40 scripts, each with its path and the digest of its content;
- 13 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
Datasets cited
- crcns.org/
data-sets/ , at CRCNS; found in the referenceshc - dandi:000054, at DANDI; found in the references
- dandi:000127, at DANDI; found in the references
- dandi:000128, at DANDI; found in the references
- dandi:000129, at DANDI; found in the references
Data availability
The new neural recordings from this study are available at Figshare (10.25378/
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
- Publisher: n/a → Nature Portfolio
Version 1, 28 September 2026: the first record
Recorded: type, language, journal, volume, issue, pages, dates, 6 authors, 2 keywords, 8 MeSH terms, 84 references.
Cite
This paper
Pachitariu, M., Zhong, L., Gracias, A., Minisi, A., Lopez, C., & Stringer, C. (2026). A critical initialization for biological neural networks. Nature, 655(8124), 990-996. https://
BibTeX
@article{pachitariu2026c
author = {Pachitariu, Marius and Zhong, Lin and Gracias, Alexa and Minisi, Amanda and Lopez, Crystall and Stringer, Carsen},
title = {{A critical initialization for biological neural networks}},
journal = {Nature},
year = {2026},
month = may,
volume = {655},
number = {8124},
pages = {990--996},
publisher = {Nature Portfolio},
issn = {0028-0836},
doi = {10.1038/
url = {https://
pmid = {42162432},
pmcid = {PMC13391357}
}
RIS
TY - JOUR
AU - Pachitariu, Marius
AU - Zhong, Lin
AU - Gracias, Alexa
AU - Minisi, Amanda
AU - Lopez, Crystall
AU - Stringer, Carsen
TI - A critical initialization for biological neural networks
T2 - Nature
J2 - Nature
PY - 2026
DA - 2026/
VL - 655
IS - 8124
SP - 990
EP - 996
SN - 0028-0836
PB - Nature Portfolio
DO - 10.1038/
UR - https://
LA - en
ER -
CSL-JSON
{
"id": "10.1038/
"type": "article-journal",
"title": "A critical initialization for biological neural networks",
"container-title": "Nature",
"author": [
{
"family": "Pachitariu",
"given": "Marius"
},
{
"family": "Zhong",
"given": "Lin"
},
{
"family": "Gracias",
"given": "Alexa"
},
{
"family": "Minisi",
"given": "Amanda"
},
{
"family": "Lopez",
"given": "Crystall"
},
{
"family": "Stringer",
"given": "Carsen"
}
],
"container-title-short":
"volume": "655",
"issue": "8124",
"page": "990-996",
"DOI": "10.1038/
"PMID": "42162432",
"PMCID": "PMC13391357",
"ISSN": "0028-0836",
"publisher": "Nature Portfolio",
"URL": "https://
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
20
]
]
}
}
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/s41593-024-01783-4 [code]
- Rastermap: a discovery method for neural population recordings.Journal: Nature neuroscienceIn common: Suite2p, Neurodata Without Borders (PyNWB, MatNWB), PyTorch, 4 other tools, mouse, 14 references, 3 authors
- [2] doi:10.1038/s41593-023-01490-6 [code]
- Facemap: a framework for modeling neural activity based on orofacial tracking.Journal: Nature neuroscienceIn common: PyTorch, scikit-learn, SciPy, 2 other tools, mouse, 12 references, 3 authors
- [3] doi:10.1038/s41586-025-09180-y [code]
- Unsupervised pretraining in biological neural networks.Journal: NatureIn common: SciPy, Matplotlib, NumPy, mouse, 9 references, 3 authors
- [4] doi:10.1038/s41467-025-61171-9 [code]
- A simplified minimodel of visual cortical neurons.Journal: Nature communicationsIn common: PyTorch, scikit-learn, SciPy, 2 other tools, mouse, 8 references, 2 authors
- [5] doi:10.1038/s41586-024-08548-w [code]
- Learning produces an orthogonalized state machine in the hippocampus.Journal: NatureIn common: Suite2p, PyTorch, scikit-learn, 3 other tools, mouse, 5 references, author Carsen Stringer
- [6] doi:10.1371/journal.pcbi.1010855 [code]
- Relating local connectivity and global dynamics in recurrent excitatory-inhibitory networks.Journal: PLoS computational biologyIn common: SciPy, Matplotlib, NumPy, 10 references
- [7] doi:10.1038/s41592-024-02232-7 [code]
- Spike sorting with Kilosort4.Journal: Nature methodsIn common: PyTorch, scikit-learn, SciPy, 2 other tools, 3 references, 2 authors
- [8] doi:10.1073/pnas.2404039122 [code]
- Structure of activity in multiregion recurrent neural networks.Journal: Proceedings of the National Academy of Sciences of the United States of AmericaIn common: PyTorch, scikit-learn, SciPy, 2 other tools, 9 references
- [9] doi:10.1038/s41586-023-06680-7 [code]
- Vagal sensory neurons mediate the Bezold-Jarisch reflex and induce syncope.Journal: NatureIn common: PyTorch, SciPy, Matplotlib, 1 other tool, 6 references, author Carsen Stringer
- [10] doi:10.1038/s41593-025-02114-x [code]
- Brain-wide analysis reveals movement encoding structured across and within brain areas.Journal: Nature neuroscienceIn common: scikit-learn, SciPy, Matplotlib, 1 other tool, mouse, 9 references
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.
Claim this paper
Correct its record
Say what each link of this record is, remove the ones that are not the paper's, add the ones that are missing. The correction becomes a new version of the record, in its Versions section.
Validate its tracing map
You validate the map as this page shows it: 2 repositories of the authors' code, each at its verified commit and with its license, 40 scripts, and 13 matches between paragraphs and code (see the Code and Map sections). It then receives a DOI on Zenodo, with you (your ORCID iD) and OSCR as its creators; the code itself is not deposited.
The map's fingerprint: sha256:6df5cc5b4a17bdeb…
Add the badge to the code's README
If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.
Markdown
<!-- oscr-badge:start -->
[](https://openscicode.org/badge/doi/10.1038/s41586-026-10528-1)
<!-- oscr-badge:end -->reStructuredText
.. oscr-badge:start
.. image:: https://openscicode.org/badge/doi/10.1038/s41586-026-10528-1.svg
:target: https://openscicode.org/badge/doi/10.1038/s41586-026-10528-1
:alt: OSCR
.. oscr-badge:endHTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.1038/s41586-026-10528-1"><img src="https://openscicode.org/badge/doi/10.1038/s41586-026-10528-1.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->On GitHub, open the README, choose Edit (the pencil), paste the snippet under the title, then "Commit changes".
If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge 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: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.
Report a mismatch between the code and the paper · Report a code error (signed in)
Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.
Report a reproduction failure (signed in)
Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.
Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.
