def plot_GAT(results,times,save_folder,compute_significance=None,suffix='SW_train_different_blocks',chance = 0.5,clim=None,tail=1):
if compute_significance is not None:
tmin_sig = compute_significance[0]
tmax_sig = compute_significance[1]
times_sig = np.where(np.logical_and(times <= tmax_sig, times > tmin_sig))[0]
sig_all = np.ones(results[0].shape)
GAT_all_for_sig = results[:, times_sig, :]
GAT_all_for_sig = GAT_all_for_sig[:, :, times_sig]
sig = stats_funcs.stats(GAT_all_for_sig-chance, tail=tail)
sig_all = SVM_funcs.replace_submatrix(sig_all, times_sig, times_sig, sig)
# -------- plot the gat --------
pretty_gat(np.mean(results,axis=0),times=times,sig=sig_all<0.05,chance = chance,clim=clim)
plt.gcf().savefig(config.fig_path+save_folder+'/'+suffix+'.png')
plt.gcf().savefig(config.fig_path+save_folder+'/'+suffix+'.svg')
plt.close('all')
def plot_results_GAT_chans_seqID(results,times,save_folder,compute_significance=None,suffix='SW_train_different_blocks',chance = 0.5,clim=None):
for chans in results.keys():
res_chan = results[chans]
for seqID in res_chan.keys():
res_chan_seq = np.asarray(res_chan[seqID])
sig_all = None
# ---- compute significance ----
if compute_significance is not None:
tmin_sig = compute_significance[0]
tmax_sig = compute_significance[1]
times_sig = np.where(np.logical_and(times <= tmax_sig, times > tmin_sig))[0]
sig_all = np.ones(res_chan_seq[0].shape)
GAT_all_for_sig = res_chan_seq[:, times_sig, :]
GAT_all_for_sig = GAT_all_for_sig[:, :, times_sig]
sig = stats_funcs.stats(GAT_all_for_sig-chance, tail=1)
sig_all = SVM_funcs.replace_submatrix(sig_all, times_sig, times_sig, sig)
# -------- plot the gat --------
pretty_gat(np.mean(res_chan_seq,axis=0),times=times,sig=sig_all<0.05,chance = 0.5,clim=clim)
plt.gcf().savefig(config.fig_path+save_folder+'/'+chans+'_'+seqID+suffix+'.png')
plt.gcf().savefig(config.fig_path+save_folder+'/'+chans+'_'+seqID+suffix+'.svg')
plt.close('all')