isolating kernels over time by cell type

visual_behavior_glm/GLM_perturbation_tools.py

@@ -10,52 +10,34 @@
 
 def analysis(weights_beh, run_params, kernel,experience_level='Familiar',savefig=False):
 
-    if kernel in ['hits','misses']:
-        lims=[[-.0005,.002],[-.01,.005]]
-    if kernel in ['omissions','all-images']:
-        lims=[[-.0005,.001],[-.015,.03]]
-
     # Plot 1D summaries
     out1 = gst.strategy_kernel_comparison(weights_beh.query('visual_strategy_session'),
         run_params, kernel,session_filter=[experience_level])
     out2 =gst.strategy_kernel_comparison(weights_beh.query('not visual_strategy_session'),
         run_params, kernel,session_filter=[experience_level])
 
-    out3 = gst.strategy_kernel_comparison(weights_beh.query('visual_strategy_session'),
-        run_params, 'all-images',session_filter=[experience_level])
-    out4 =gst.strategy_kernel_comparison(weights_beh.query('not visual_strategy_session'),
-        run_params, 'all-images',session_filter=[experience_level])
-
     # Unify ylimits and add time markers
     ylim1 = out1[2].get_ylim()
     ylim2 = out2[2].get_ylim()
-    ylim3 = out3[2].get_ylim()
-    ylim4 = out4[2].get_ylim()
-    ylim = [np.min([ylim1[0],ylim2[0],ylim3[0],ylim4[0]]), \
-        np.max([ylim1[1],ylim2[1],ylim3[1],ylim4[1]])]
+    ylim = [np.min([ylim1[0],ylim2[0]]), \
+        np.max([ylim1[1],ylim2[1]])]
 
     out1[2].set_ylim(ylim)
     out2[2].set_ylim(ylim)
-    out3[2].set_ylim(ylim)
-    out4[2].set_ylim(ylim)
     out1[2].set_ylabel(kernel+' kernel\n(Ca$^{2+}$ events)',fontsize=16)
     out2[2].set_ylabel(kernel+' kernel\n(Ca$^{2+}$ events)',fontsize=16)
-    out3[2].set_ylabel('image kernel\n(Ca$^{2+}$ events)',fontsize=16)
-    out4[2].set_ylabel('image kernel\n(Ca$^{2+}$ events)',fontsize=16)
-
     out1[2].set_title('Visual strategy',fontsize=16)
     out2[2].set_title('Timing strategy',fontsize=16)
-    out3[2].set_title('Visual strategy',fontsize=16)
-    out4[2].set_title('Timing strategy',fontsize=16)
+
     if kernel =='omissions':
         out1[2].plot(0,ylim[0],'co',zorder=10,clip_on=False)
         out2[2].plot(0,ylim[0],'co',zorder=10,clip_on=False)
     elif kernel =='hits':
         out1[2].plot(0,ylim[0],'ro',zorder=10,clip_on=False)
         out2[2].plot(0,ylim[0],'ro',zorder=10,clip_on=False)
     elif kernel == 'misses':
-        out1[2].plot(0,ylim[0],'rx',zorder=10,clip_on=False)
-        out2[2].plot(0,ylim[0],'rx',zorder=10,clip_on=False)
+        out1[2].plot(0,ylim[0],'ro',zorder=10,clip_on=False)
+        out2[2].plot(0,ylim[0],'ro',zorder=10,clip_on=False)
     else:
         out1[2].plot(0,ylim[0],'ko',zorder=10,clip_on=False)
         out2[2].plot(0,ylim[0],'ko',zorder=10,clip_on=False)
@@ -65,10 +47,6 @@ def analysis(weights_beh, run_params, kernel,experience_level='Familiar',savefig
         out2[2].plot(.75,ylim[0],'ko',zorder=10,clip_on=False)
         out2[2].plot(1.5,ylim[0],'o',color='gray',zorder=10,clip_on=False)
 
-    # Make 2D plot
-    ax = plot_perturbation(weights_beh, run_params, kernel,experience_level=experience_level,
-        savefig=savefig,lims=lims)
-
     if savefig:
         filepath = run_params['figure_dir']+\
             '/strategy/'+kernel+'_visual_comparison_{}.svg'.format(experience_level)
@@ -80,14 +58,6 @@ def analysis(weights_beh, run_params, kernel,experience_level='Familiar',savefig
         out2[1].savefig(filepath) 
         filepath = run_params['figure_dir']+\
             '/strategy/images_visual_comparison_{}.svg'.format(experience_level)
-        print('Figure saved to: '+filepath)
-        out3[1].savefig(filepath) 
-        filepath = run_params['figure_dir']+\
-            '/strategy/images_timing_comparison_{}.svg'.format(experience_level)
-        print('Figure saved to: '+filepath)
-        out4[1].savefig(filepath) 
-
-    return ax 
 
 def get_kernel_averages(weights_df, run_params, kernel, drop_threshold=0,
     session_filter=['Familiar','Novel 1','Novel >1'],equipment_filter="all",

visual_behavior_glm/strategy_paper_figure_script.py

@@ -140,7 +140,14 @@
 GLM_VERSION = '24_events_all_L2_optimize_by_session'
 run_params, results, results_pivoted, weights_df = gfd.get_analysis_dfs(GLM_VERSION)
 weights_beh = gst.add_behavior_session_metrics(weights_df, summary_df)
+
+# Plot kernels over time, compare cell types
+gpt.analysis(weights_beh, run_params, 'all-images')
 gpt.analysis(weights_beh, run_params, 'omissions')
+gpt.analysis(weights_beh, run_params, 'hits')
+gpt.analysis(weights_beh, run_params, 'misses')
+
+# Plot kernels over time, compare strategies
 gst.kernels_by_cre(weights_beh, run_params)