Figs correction (#157)

* Replace polish names

* Uncomment data path

* Modify make_plot_perf

to take noise_lvl as an argument (#150)

* Update CSD resolution

* Replace deprecated griddata function

* Update README to figures

* Fix figures

* Update README

* Fix data shapes in figures

---------

Co-authored-by: Marta <[email protected]>

Author: m-kowalska

figures/kCSD_properties/L_curve_simulation.py

@@ -118,7 +118,7 @@ def main_loop(src_width, total_ele, inpos, lpos, nm, noise=0, srcs=1):
     for i, noise in enumerate(n_spec):
         plt.close('all')
         noise = np.round(noise, 5)
-        print('numer rekonstrukcji: ', i, 'noise level: ', noise)
+        print('number of reconstruction: ', i, 'noise level: ', noise)
         #Electrodes
         ele_pos, pots = electrode_config(total_ele, true_csd, t_csd_x, t_csd_y, inpos, lpos)
         ele_y = ele_pos[:, 1]
@@ -179,5 +179,5 @@ def main_loop(src_width, total_ele, inpos, lpos, nm, noise=0, srcs=1):
         os.chdir('..')
     np.save('sim_results', sim_results)
     sim_results = np.load('sim_results.npy')
-    make_plot_perf(sim_results)
+    make_plot_perf(sim_results, noise_lvl)
 

figures/kCSD_properties/README.txt

@@ -1,178 +1,93 @@
-Instructions for the figures from kCSD-revisited paper.
+Instructions for the figures from kCSD-python, reliable current source density estimation with quality control.
 
 ~~~~~~~~~~~~~~~~~~~~~~~
 Figure 1 - Schematic
 
 name: figure1.png
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 2 - 1D spectral properties of kCSD method
-
-figure_eigensources_M_1D.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 3 - Noise-free Electrode / Basis source placement
-
-figure_Tbasis.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 4 - Noisy electrodes / Basis source placement
-
-figure_Tbasis_noise.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 5 - L-curve method for regularization
+Figure 2 - L-curve method for regularization
 
 You will need to run L_curve_simulation.py first.
 
 figure_LC.py
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 6 - L-curve versus Cross-validation
-
-You will need to run L_curve_simulation.py first.
-
-figure_LCandCVperformance.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 7 - Error propagation map
+Figure 3 - Error propagation map
 
 error_propagation.py
 colorblind_friendly.py
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 8 - Reliability map
+Figure 4 - Reliability map 2D
 
 reliability_map_2d.py
 
 error_maps_2D/point_error_large_100_all_ele.npy
 error_maps_2D/point_error_small_100_all_ele.npy
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 9 - Reliability map; Use case in a 2D dipolar large source
+Figure 5 - Reliability map; Use case in a 2D dipolar large source
 
 kCSD_with_reliability_map_2D.py
 
 error_maps_2D/point_error_large_100_all_ele.npy
 error_maps_2D/point_error_small_100_all_ele.npy
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 10 - Average Error (Diff) when broken electrode and loss in reconstruction quality
+Figure 6 - Average Error (Diff) when broken electrode and loss in reconstruction quality
 
 You will need to run tutorial3.py first or download files from here
 https://www.dropbox.com/sh/6kykj4d3dx3fp5s/AAACtN49VCbAHA9otOfNXbnOa?dl=0
 
 tutorial_broken_electrodes_diff_err.py
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 11 - Simulated cortical recordings in Traubs's model
-
-You will need to download files from:
-https://repod.pon.edu.pl/dataset/thalamocortical-network/resource/6add09e1-bfe4-4082-b990-24b469756886
-
-npx/traub_data_kcsd_column_figure.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 12 - LFP and CSD as a function of time - Traub's model
-
-You will need to download files from:
-https://repod.pon.edu.pl/dataset/thalamocortical-network/resource/6add09e1-bfe4-4082-b990-24b469756886
-
-npx/figure_traub_timespace.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 13 - Six first eigensources for a single bank of a Neuropixels probe
-
-You will need to download files from:
-https://repod.pon.edu.pl/dataset/thalamocortical-network/resource/6add09e1-bfe4-4082-b990-24b469756886
-
-npx/figure_traub_eigensources.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 14 - LFP and CSD space profiles - experimental data from a single bank of a Neuropixels probe
 
-You will need to download files from:
-
-npx/kCSD2D_reconstruction_from_npx.py
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Figure 15 - LFP and CSD as a function of time - experimental data
-
-You will need to download files from:
-
-npx/kCSD2D_reconstruction_from_npx.py
-
-~~~~~~~~~~~~~~~~~~~~~~~~
-Figure 16 - L-Curve and CV landscape
+Figure 7 - L-Curve and CV landscape
 
 You will need to run L_curve_simulation.py first.
 
 figure_LCandCV.py
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Figure 17 - Schematic - location of Neuropixels bank 0
-
-name: figure17.png
-
-=====================
-Supplementary Figures
-=====================
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~
-Figure 2-Figure supplement 1 - spectral properties of kCSD method for simple 2D case
-
-figure_eigensources_M_2D.py
-
-~~~~~~~~~~~~~~~~~~~~~~~~~~
-Figure 7-Figure supplement 1 - Error propagation maps for 1D
-
-pots_propagation.py
 
-~~~~~~~~~~~~~~~~~~~~~~~~~~
-Figure 13-Figure supplement 1 - Eigensurces 7-12 ('start=6', 'stop=12'),
-Figure 13-Figure supplement 2 - Eigensurces 13-18 ('start=12', 'stop=18'),
-Figure 13-Figure supplement 3 - Eigensurces 19-24 ('start=18', 'stop=24'),
-Figure 13-Figure supplement 4 - Eigensurces 25-30 ('start=24', 'stop=30'),
-Figure 13-Figure supplement 5 - Eigensurces 31-36 ('start=30', 'stop=36'),
-Figure 13-Figure supplement 6 - Eigensurces 37-42 ('start=36', 'stop=42'),
-Figure 13-Figure supplement 7 - Eigensurces 43-48 ('start=42', 'stop=48')
-
-All supplementary figures to Figure 13 were created using different
-'start' and 'stop'parameters at:
-npx/figure_traub_eigensources.py
-
-================
-Appendix Figures
-================
-
-~~~~~~~~~~~~~~~~~~~~~~~
-Appendix 1 Figure 1 - Basic features tutorial
+Figure 8 - Basic features tutorial
 
 You will need to run tutorial3.py first or download files from here
 https://www.dropbox.com/sh/6kykj4d3dx3fp5s/AAACtN49VCbAHA9otOfNXbnOa?dl=0
 
 tutorial_basic.py
 
 ~~~~~~~~~~~~~~~~~~~~~~~
-Appendix 1 Figure 2 - Noisy electrodes tutorial
+
+Figure 9 - Noisy electrodes tutorial
 
 tutorial_noisy_electrodes.py
 
-~~~~~~~~~~~~~~~~~~~~~~~~
-Appendix 1 Figure 3 - Broken electrodes tutorial
+~~~~~~~~~~~~~~~~~~~~~~~
+Figure 10 - Broken electrodes tutorial
 
 Download first from
 https://www.dropbox.com/sh/6kykj4d3dx3fp5s/AAACtN49VCbAHA9otOfNXbnOa?dl=0
 (generated from tweaking tutorial3.py)
 
 tutorial_broken_electrodes.py
 
-~~~~~~~~~~~~~~~~~~~~~~~~~~
-Appendix 1 Figure 4 - 3D source reconstruction
+~~~~~~~~~~~~~~~~~~~~~~~
+Figure 11 - Error propagation maps for 1D
+
+pots_propagation.py
+
+~~~~~~~~~~~~~~~~~~~~~~~
+Figure 12 - 3D source reconstruction
 
 tutorial_basic_3d.py
 
-~~~~~~~~~~~~~~~~~~~~~~~~~~
-Appendix 1 Figure 5 - sKCSD example
+~~~~~~~~~~~~~~~~~~~~~~~
+Figure 13 - sKCSD example
+
+You will need to install LFPy package first:
+pip install lfpy
 
 skcsd_and_l_curve_complex_morphology.py

figures/kCSD_properties/error_propagation.py

@@ -80,8 +80,8 @@ def _BipolarColormap(name, negative, positive):
 h = 50. # distance between the electrode plane and the CSD plane
 conductivity = 1.0 # S/m
 
-csd_at = np.mgrid[0.:1.:100j,
-                  0.:1.:100j]
+csd_at = np.mgrid[0.:1.:101j,
+                  0.:1.:101j]
 csd_x, csd_y = csd_at
 
 D = 2 - 1.618

figures/kCSD_properties/figure_LC.py

@@ -100,12 +100,12 @@ def make_plots(title, m_norm, m_resi, true_csd, curveseq, ele_y,
     fig.savefig(save_as+'.png')
 
 if __name__=='__main__':
-#    os.chdir("./LCurve/LC2")
+    os.chdir("./LCurve/LC2")
     noises = 3
     noise_lvl = np.linspace(0, 0.5, noises)
-    #df = np.load('data_fig4_and_fig13_lc_noise25.0.npz')
+    df = np.load('data_fig4_and_fig13_LC_noise25.0.npz')
     Rs = np.linspace(0.025, 8*0.025, 8)
-    title = ['nazwa_pliku']
+    title = ['file_name']
     save_as = 'noise'
     make_plots(title, df['m_norm'], df['m_resi'], df['true_csd'], 
                df['curve_surf'], df['ele_y'], df['pots_n'],

figures/kCSD_properties/figure_LCandCV.py

@@ -74,7 +74,7 @@ def plot_surface(curve_surf, errsy, save_as):
     df = np.load(os.path.join('LC2', 'data_fig4_and_fig13_LC_noise25.0.npz'))
 
     Rs = np.linspace(0.025, 8*0.025, 8)
-    title = ['nazwa_pliku']
+    title = ['file_name']
     save_as = 'noise'
     plot_surface(df['curve_surf'], df['errsy'], save_as+'surf')
     plt.close('all')

figures/kCSD_properties/figure_LCandCVperformance.py

@@ -20,7 +20,7 @@ def set_axis(ax, x, y, letter=None):
             transform=ax.transAxes)
     return ax
 
-def make_plot_perf(sim_results):
+def make_plot_perf(sim_results, noise_lvl):
     rms_lc = sim_results[0, 2]
     lam_lc = sim_results[0, 0]
     rms_cv = sim_results[1, 2]
@@ -81,4 +81,4 @@ def make_plot_perf(sim_results):
     noises = 9
     noise_lvl = np.linspace(0, 0.5, noises)
     sim_results = np.load('sim_results.npy')
-    make_plot_perf(sim_results)
+    make_plot_perf(sim_results, noise_lvl)

figures/kCSD_properties/kCSD_with_reliability_map_2D.py

@@ -102,7 +102,7 @@ def make_reconstruction(KK, csd_profile, csd_seed, total_ele,
 
 def make_subplot(ax, val_type, xs, ys, values, cax, title=None, ele_pos=None,
                  xlabel=False, ylabel=False, letter='', t_max=None,
-                 mask=False, level=False):
+                 mask=False, mask_x=False, mask_y=False, level=False):
     if val_type == 'csd':
         cmap = cm.bwr
     elif val_type == 'pot':
@@ -124,7 +124,10 @@ def make_subplot(ax, val_type, xs, ys, values, cax, title=None, ele_pos=None,
                          levels=levels, cmap=cmap, alpha=1,
                          extent=(0, 0.5, 0, 0.5))
     if mask is not False:
-        CS = ax.contour(xs, ys, mask, cmap='Greys')
+        if mask_x is not False:
+            CS = ax.contour(mask_x, mask_y, mask, cmap='Greys')
+        else:
+            CS = ax.contour(xs, ys, mask, cmap='Greys')
         ax.clabel(CS,  # label every second level
                   inline=1,
                   fmt='%1.2f',
@@ -182,7 +185,7 @@ def generate_figure(k, true_csd, ele_pos, pots, mask=False):
 #    gs.update(top=.95, bottom=0.53)
     ax = plt.subplot(gs[0, 0])
     cax = plt.subplot(gs[1, 0])
-    make_subplot(ax, 'csd', csd_x, csd_y, true_csd, cax=cax, ele_pos=ele_pos,
+    make_subplot(ax, 'csd', k.estm_x, k.estm_y, true_csd, cax=cax, ele_pos=ele_pos,
                  title='True CSD', xlabel=True, ylabel=True, letter='A',
                  t_max=np.max(abs(true_csd)))
     ax = plt.subplot(gs[0, 1])
@@ -195,10 +198,10 @@ def generate_figure(k, true_csd, ele_pos, pots, mask=False):
     make_subplot(ax, 'csd', k.estm_x, k.estm_y, k.values('CSD')[:, :, 0],
                  cax=cax, ele_pos=ele_pos, title='kCSD with Reliability Map',
                  xlabel=True, letter='C', t_max=np.max(abs(true_csd)),
-                 mask=mask)
+                 mask=mask, mask_x=csd_x, mask_y=csd_y)
     ax = plt.subplot(gs[0, 3])
     cax = plt.subplot(gs[1, 3])
-    make_subplot(ax, 'diff', csd_x, csd_y,
+    make_subplot(ax, 'diff', k.estm_x, k.estm_y,
                  abs(true_csd-k.values('CSD')[:, :, 0]), cax=cax,
                  ele_pos=ele_pos, title='|True CSD - kCSD|', xlabel=True,
                  letter='D',
@@ -241,9 +244,10 @@ def matrix_symmetrization(point_error):
                                                              Rs=Rs,
                                                              lambdas=lambdas,
                                                              method=method)
+    test_csd = CSD_PROFILE([k.estm_x, k.estm_y], CSD_SEED)
     error_l = np.load('error_maps_2D/point_error_large_100_all_ele.npy')
     error_s = np.load('error_maps_2D/point_error_small_100_all_ele.npy')
     error_all = np.concatenate((error_l, error_s))
     symm_array_all = matrix_symmetrization(error_all)
-    generate_figure(k, true_csd, ele_pos, pots,
+    generate_figure(k, test_csd, ele_pos, pots,
                     mask=np.mean(symm_array_all, axis=0))

figures/kCSD_properties/pots_propagation.py

@@ -123,11 +123,11 @@ def generate_figure(true_csd_xlims, total_ele, ele_lims, R_init=0.23):
     for i in range(len(ele_pos)):
         ax = fig.add_subplot(gs[plt_cord[i][0],
                                 plt_cord[i][1]:plt_cord[i][1]+2])
-        ax.plot(np.linspace(0, 1, 100), est_csd[i], lw=2, c='red',
+        ax.plot(np.linspace(0, 1, 101), est_csd[i], lw=2, c='red',
                 label='kCSD')
         ax.scatter(ele_pos, np.zeros(len(ele_pos)), c='k', label='Electrodes')
         ax2 = ax.twinx()
-        ax2.plot(np.linspace(0, 1, 100), OBJ_M[i].values('POT'), c='green',
+        ax2.plot(np.linspace(0, 1, 101), OBJ_M[i].values('POT'), c='green',
                  label='Potentials')
         ax2.set_ylim([-1.5, 1.5])
         ax.set_ylim([-150, 150])

figures/kCSD_properties/reliability_map_2D.py

@@ -73,7 +73,7 @@ def make_single_subplot(ax, val_type, xs, ys, values, cax, title=None,
     plt.colorbar(im, cax=cax, orientation='horizontal', format='%.2f',
                  ticks=ticks)
     set_axis(ax, letter=letter)
-    plt.tight_layout()
+    #plt.tight_layout()
     return ax, cax