fixes

Author: mmoskon

CosinorPy/cosinor.py

@@ -502,7 +502,7 @@ def population_fit(df_pop, n_components = 2, period = 24, lin_comp= False, model
 
 
         #results, statistics, rhythm_params, X_test, Y_test, model = fit_me(x, y, n_components = n_components, period = period, lin_comp=lin_comp, model_type = model_type, alpha=alpha, plot = False, return_model = True)
-        results, statistics, rhythm_params, X_test, Y_test, model = fit_me(x, y, n_components = n_components, period = period, plot = False, return_model = True, **kwargs)
+        results, statistics, rhythm_params, X_test, _, model = fit_me(x, y, n_components = n_components, period = period, plot = False, return_model = True, **kwargs)
         if type(params) == int:
             params = results.params
         else:
@@ -577,7 +577,7 @@ def population_fit(df_pop, n_components = 2, period = 24, lin_comp= False, model
 
 
     if plot_on and plot_margins and model_type=='lin':
-        sdev, lower, upper = wls_prediction_std(results, exog=X_fit_eval_params, alpha=0.05)
+        _, lower, upper = wls_prediction_std(results, exog=X_fit_eval_params, alpha=0.05)
         plt.fill_between(X_test, lower, upper, color='#888888', alpha=0.1)                   
 
     if plot_measurements:
@@ -634,7 +634,7 @@ def generate_permutations(n_pop1, n_pop2, N):
 """
 def generate_permutations_all(pop1, pop2):
     n_pop1 = len(pop1)
-    n_pop2 = len(pop2)
+    #n_pop2 = len(pop2)
 
     permutations = set()
 
@@ -696,13 +696,10 @@ def permutation_test_population(df, pairs, period = 24, n_components = 2, lin_co
 
         tests1 = list(df_pop1.test.unique())
         tests2 = list(df_pop2.test.unique())
-        n_pop1 = len(tests1)
-        n_pop2 = len(tests2)
+        #n_pop1 = len(tests1)
+        #n_pop2 = len(tests2)
 
-        #print(tests1)
-        #print(tests2)
-
-        tests = np.array(tests1 + tests2)
+        #tests = np.array(tests1 + tests2)
 
         """
         if not permutations:            
@@ -881,7 +878,7 @@ def fit_me(X, Y, n_components = 2, period = 24, lin_comp = False, model_type = '
     if plot:
         if plot_margins:
             if model_type == 'lin':
-                sdev, lower, upper = wls_prediction_std(results, exog=X_fit_test, alpha=0.05)
+                _, lower, upper = wls_prediction_std(results, exog=X_fit_test, alpha=0.05)
                 if color:
                     plt.fill_between(X_test, lower, upper, color=color, alpha=0.1)
                 else:
@@ -1019,7 +1016,7 @@ def fit_me(X, Y, n_components = 2, period = 24, lin_comp = False, model_type = '
                 plt.show()
             if plot_residuals:
                 resid = results.resid
-                fig = sm.qqplot(resid)
+                sm.qqplot(resid)
                 plt.title(name)
                 if save_to:
                     plt.savefig(save_to+'_resid.pdf', bbox_inches='tight')
@@ -1076,7 +1073,7 @@ def plot_phases(acrs, amps, tests, period=24, colors = ("black", "red", "green",
     ax.set_theta_direction(-1) 
     lines = []
 
-    for i, (acr, amp, test, color) in enumerate(zip(acrs, amps, tests, colors)):
+    for i, (acr, amp, _, color) in enumerate(zip(acrs, amps, tests, colors)):
 
         """
         if "LDL" in test:
@@ -1336,7 +1333,7 @@ def compare_pairs(df, pairs, n_components = 3, period = 24, folder = "", prefix
                     save_to = ''
 
                 #pvalues, params, results = compare_pair_df_extended(df, test1, test2, n_components = n_comps, period = per, lin_comp = lin_comp, model_type = model_type, alpha=alpha, save_to = save_to, plot_measurements=plot_measurements)
-                pvalues, params, results = compare_pair_df_extended(df, test1, test2, n_components = n_comps, period = per, save_to = save_to, **kwargs)
+                pvalues, params, _ = compare_pair_df_extended(df, test1, test2, n_components = n_comps, period = per, save_to = save_to, **kwargs)
 
                 d = {}
                 d['test'] = test1 + ' vs. ' + test2
@@ -1370,11 +1367,6 @@ def compare_pairs(df, pairs, n_components = 3, period = 24, folder = "", prefix
 
     #return multi.multipletests(P, method = 'fdr_bh')[1]
 
-
-
-
-
-
 #def compare_pairs_best_models(df, df_best_models, pairs, lin_comp = False, model_type = 'lin', alpha = 0, folder = '', prefix = '', plot_measurements=True):
 def compare_pairs_best_models(df, df_best_models, pairs, folder = "", prefix = "", **kwargs):
     df_results = pd.DataFrame()
@@ -1402,7 +1394,7 @@ def compare_pairs_best_models(df, df_best_models, pairs, folder = "", prefix = "
             save_to = ''
 
         #pvalues, params, results = compare_pair_df_extended(df, test1, test2, n_components = n_components1, period = period1, n_components2 = n_components2, period2 = period2, lin_comp = lin_comp, model_type = model_type, alpha=alpha, save_to = save_to, plot_measurements=plot_measurements)
-        pvalues, params, results = compare_pair_df_extended(df, test1, test2, n_components = n_components1, period = period1, n_components2 = n_components2, period2 = period2, **kwargs)
+        pvalues, params, _ = compare_pair_df_extended(df, test1, test2, n_components = n_components1, period = period1, n_components2 = n_components2, period2 = period2, save_to = save_to, **kwargs)
 
         d = {}
         d['test'] = test1 + ' vs. ' + test2
@@ -1633,9 +1625,9 @@ def compare_pair_df_extended(df, test1, test2, n_components = 3, period = 24, n_
     plt.plot(X_full, Y_fit_full2, 'r', label = test2)    
 
     if model_type == 'lin' and plot_margins:
-        sdev, lower, upper = wls_prediction_std(results, exog=X_fit_full[locs], alpha=0.05)
+        _, lower, upper = wls_prediction_std(results, exog=X_fit_full[locs], alpha=0.05)
         plt.fill_between(X_full, lower, upper, color='black', alpha=0.1)   
-        sdev, lower, upper = wls_prediction_std(results, exog=X_fit_full[~locs], alpha=0.05)
+        _, lower, upper = wls_prediction_std(results, exog=X_fit_full[~locs], alpha=0.05)
         plt.fill_between(X_full, lower, upper, color='red', alpha=0.1)
 
 
@@ -1662,7 +1654,7 @@ def compare_pair_df_extended(df, test1, test2, n_components = 3, period = 24, n_
     if plot_residuals:
 
         resid = results.resid
-        fig = sm.qqplot(resid)
+        sm.qqplot(resid)
         plt.title(test1 + ' vs. ' + test2)
         save_to_resid = save_to.split(".")[0] + '_resid' + save_to.split(".")[1]
         if save_to:
@@ -1852,7 +1844,7 @@ def compare_pair(X1, Y1, X2, Y2, test1 = '', test2 = '', n_components = 3, perio
     if plot_residuals:
 
         resid = results.resid
-        fig = sm.qqplot(resid)
+        sm.qqplot(resid)
         plt.title(test1 + ' vs. ' + test2)
         save_to_resid = save_to.split(".")[0] + '_resid' + save_to.split(".")[1]
         if save_to:
@@ -2086,7 +2078,7 @@ def compare_pair_extended(X1, Y1, X2, Y2, test1 = '', test2 = '', n_components1
     if plot_residuals:
 
         resid = results.resid
-        fig = sm.qqplot(resid)
+        sm.qqplot(resid)
         plt.title(test1 + ' vs. ' + test2)
         save_to_resid = save_to.split(".")[0] + '_resid' + save_to.split(".")[1]
         if save_to:
@@ -2519,8 +2511,8 @@ def compare_nonlinear(X1, Y1, X2, Y2, test1 = '', test2 = '', min_per = 18, max_
     plt.plot(X1, Y1, 'ko', markersize=1, label = test1)
     plt.plot(X2, Y2, 'ro', markersize=1, label = test2)
 
-    Y_fit1 = Y_fit[H == 0]
-    Y_fit2 = Y_fit[H == 1]
+    #Y_fit1 = Y_fit[H == 0]
+    #Y_fit2 = Y_fit[H == 1]
 
     plt.plot(X_full, Y1_full, 'k', label = 'fit '+test1)    
     plt.plot(X_full, Y2_full, 'r', label = 'fit '+test2)    
@@ -2542,7 +2534,7 @@ def compare_nonlinear(X1, Y1, X2, Y2, test1 = '', test2 = '', min_per = 18, max_
     if plot_residuals:
 
         resid = Y-Y_fit
-        fig = sm.qqplot(resid)
+        sm.qqplot(resid)
         plt.title(test1 + ' vs. ' + test2)
         save_to_resid = save_to + '_resid' 
         if save_to:

CosinorPy/cosinor1.py

@@ -74,12 +74,12 @@ def plot_pair(data, results, test1='1', test2='2', plot_measurements=True, save_
 
             D1 = np.column_stack((H1_fit, rrr_fit_M, sss_fit_M, H1_fit*rrr_fit_M, H1_fit*sss_fit_M))
             D1 = sm.add_constant(D1, has_constant='add')
-            sdev, lower, upper = wls_prediction_std(results, exog=D1, alpha=0.05)
+            _, lower, upper = wls_prediction_std(results, exog=D1, alpha=0.05)
             plt.fill_between(X_fit_M, lower, upper, color=color1, alpha=0.1)   
 
             D2 = np.column_stack((H2_fit, rrr_fit_M, sss_fit_M, H2_fit*rrr_fit_M, H2_fit*sss_fit_M))
             D2 = sm.add_constant(D2, has_constant='add')
-            sdev, lower, upper = wls_prediction_std(results, exog=D2, alpha=0.05)
+            _, lower, upper = wls_prediction_std(results, exog=D2, alpha=0.05)
             plt.fill_between(X_fit_M, lower, upper, color=color2, alpha=0.1)   
 
 
@@ -157,7 +157,7 @@ def plot_single(data, results, test='', plot_measurements=True, save_to='', plot
         D = sm.add_constant(D, has_constant='add')
 
         if plot_margins:
-            sdev, lower, upper = wls_prediction_std(results, exog=D, alpha=0.05)
+            _, lower, upper = wls_prediction_std(results, exog=D, alpha=0.05)
             plt.fill_between(X_fit, lower, upper, color=color, alpha=0.1)   
             #plt.fill_between(X_fit, lower, upper, color='#888888', alpha=0.1)   
 
@@ -402,7 +402,7 @@ def population_fit_cosinor(df_pop, period, save_to='', alpha = 0.05, plot_on = T
 
     for test in tests:
         x,y = np.array(df_pop[df_pop.test == test].x), np.array(df_pop[df_pop.test == test].y)
-        fit_results, amp, acr, statistics = fit_cosinor(x, y, period = period, save_to=save_to, plot_on = False)
+        fit_results, amp, acr, _ = fit_cosinor(x, y, period = period, save_to=save_to, plot_on = False)
         if plot_on and plot_individuals:
             X_fit = np.linspace(min(x), max(x), 100)
             rrr_fit= np.cos(2*np.pi*X_fit/period)
@@ -555,27 +555,27 @@ def population_test_cosinor(pop1, pop2):
     mesors2 = params2[:,0]
     mesor2 = means2[0]
     betas1 = params1[:,1]
-    beta1 = means1[1]
+    #beta1 = means1[1]
     betas2 = params2[:,1]
-    beta2 = means2[1]
+    #beta2 = means2[1]
     gammas1 = params1[:,2]
-    gamma1 = means1[2]
+    #gamma1 = means1[2]
     gammas2 = params2[:,2]
-    gamma2 = means2[2]
-    amps1 = params1[:,3]
+    #gamma2 = means2[2]
+    #amps1 = params1[:,3]
     amp1 = means1[3]
-    amps2 = params2[:,3]
+    #amps2 = params2[:,3]
     amp2 = means2[3]
-    acrs1 = params1[:,4]
+    #acrs1 = params1[:,4]
     acr1 = means1[4]
-    acrs2 = params2[:,4]
+    #acrs2 = params2[:,4]
     acr2 = means2[4]
 
     M = (k1*mesor1 + k2*mesor2)/K
     A = (k1*amp1 + k2*amp2)/K
     FI = (k1*acr1 + k2*acr2)/K
-    BETA = (k1*beta1 + k2*beta2)/K
-    GAMMA = (k1*gamma1 + k2*gamma2)/K
+    #BETA = (k1*beta1 + k2*beta2)/K
+    #GAMMA = (k1*gamma1 + k2*gamma2)/K
     TM = (k1 * (mesor1 - M)**2) + (k2 * (mesor2 - M)**2)
 
 
@@ -719,10 +719,10 @@ def test_cosinor_pairs(df, pairs, period = 24, folder = '', prefix='', plot_meas
         idx_group_acr = 5
 
         if type(df_best_models) == int:
-            fit_results, statistics_raw, statistics_trans, global_test_amp, ind_test_amp, global_test_acr, ind_test_acr = test_cosinor_pair(df_pair, period)
+            fit_results, _, statistics_trans, _, ind_test_amp, _, ind_test_acr = test_cosinor_pair(df_pair, period)
         else:            
             period = df_best_models[df_best_models.test == test1].period.iloc[0]
-            fit_results, statistics_raw, statistics_trans, global_test_amp, ind_test_amp, global_test_acr, ind_test_acr = test_cosinor_pair(df_pair, period)
+            fit_results, _, statistics_trans, _, ind_test_amp, _, ind_test_acr = test_cosinor_pair(df_pair, period)
             #print(period)
 
         acr1, acr2 = statistics_trans['values'][idx_acr], statistics_trans['values'][idx_group_acr]