gwaslab 3.4.37__py3-none-any.whl → 3.4.39__py3-none-any.whl

gwaslab/viz_plot_trumpetplot.py

@@ -66,7 +66,7 @@ def plottrumpet(mysumstats,
                 anno_source = "ensembl",
                 anno_max_iter=100,
                 arm_scale=1,
-                repel_force=0.05,
+                repel_force=0.01,
                 ylabel="Effect size",
                 xlabel="Minor allele frequency",
                 xticks = None,
@@ -99,7 +99,7 @@ def plottrumpet(mysumstats,
             xticks = [0,0.01,0.05,0.1,0.2,0.5]
             xticklabels = xticks            
     if figargs is None:
-        figargs={"figsize":(10,15)}
+        figargs={"figsize":(10,8)}
     if scatter_args is None:
         scatter_args ={}
     if hue is not None:
@@ -108,36 +108,34 @@ def plottrumpet(mysumstats,
         scatter_args["color"]=markercolor
     
     #Checking columns#################################################################################################################
-    if verbose: log.write("Start to create trumpet plot...")
+    log.write("Start to create trumpet plot...", verbose=verbose)
     
     #parameter check##################################################################################################################
     if (beta not in mysumstats.columns) or (eaf not in mysumstats.columns):
-        if verbose:
-            log.write(" -No EAF or BETA columns. Skipping...")
+        log.write(" -No EAF or BETA columns. Skipping...", verbose=verbose)
         return None
     if mode=="b":
         if ncase is None or ncontrol is None:
-            if verbose:
-                log.write(" -No scase or scontrol. Skipping...")
+            log.write(" -No scase or scontrol. Skipping...", verbose=verbose)
             return None
         if prevalence is None:
                 prevalence= ncase/(ncase + ncontrol)
-                log.write(" -Prevalence is not given. Estimating based on scase and scontrol :{}...".format(prevalence))
+                log.write(" -Prevalence is not given. Estimating based on scase and scontrol :{}...".format(prevalence), verbose=verbose)
     
     #print settings##################################################################################################################
-    if verbose:
-        log.write(" -Settings:")
-        log.write("  -Mode: {}".format(mode))
-        if mode == "q" :
-            log.write("  -N: {}".format(n))
-        if mode == "b" :
-            log.write("  -N_CASE: {}".format(ncase))
-            log.write("  -N_CONTROL: {}".format(ncontrol))
-            log.write("  -PREVALENCE: {}".format(prevalence))
-        log.write("  -BETA: {}".format(beta))
-        log.write("  -Significance level: {}".format(sig_level))
-        log.write("  -Power thresholds: {}".format(ts))
-        log.write("  -Power line smoothness: {}".format(n_matrix))
+
+    log.write(" -Settings:", verbose=verbose)
+    log.write("  -Mode: {}".format(mode), verbose=verbose)
+    if mode == "q" :
+        log.write("  -N: {}".format(n), verbose=verbose)
+    if mode == "b" :
+        log.write("  -N_CASE: {}".format(ncase), verbose=verbose)
+        log.write("  -N_CONTROL: {}".format(ncontrol), verbose=verbose)
+        log.write("  -PREVALENCE: {}".format(prevalence), verbose=verbose)
+    log.write("  -BETA: {}".format(beta), verbose=verbose)
+    log.write("  -Significance level: {}".format(sig_level), verbose=verbose)
+    log.write("  -Power thresholds: {}".format(ts), verbose=verbose)
+    log.write("  -Power line smoothness: {}".format(n_matrix), verbose=verbose)
     
     #loading columns #################################################################################################################
     cols_to_use = [snpid, beta, eaf, n, p]
@@ -145,7 +143,7 @@ def plottrumpet(mysumstats,
     if anno is not None:
         if anno != "GENENAME":
             if anno!=True:
-                log.write(" -Loading column {} for annotation...".format(anno))
+                log.write(" -Loading column {} for annotation...".format(anno), verbose=verbose)
                 if anno not in cols_to_use:
                     cols_to_use.append(anno)
         else:
@@ -160,17 +158,17 @@ def plottrumpet(mysumstats,
     #filter by p #################################################################################################################
     if p in mysumstats.columns:
         sumstats = mysumstats.loc[mysumstats[p]< p_level,cols_to_use ].copy()
-        if verbose: log.write(" -Excluding variants with P values > {}".format(p_level))
+        log.write(" -Excluding variants with P values > {}".format(p_level), verbose=verbose)
     else:
         cols_to_use.remove(p)
         sumstats = mysumstats[[beta,eaf,n]].copy()
-    if verbose: log.write(" -Plotting {} variants...".format(len(sumstats)))
+    log.write(" -Plotting {} variants...".format(len(sumstats)), verbose=verbose)
     
     #add maf column #################################################################################################################
     if maf not in sumstats.columns:
         sumstats = filldata(sumstats,to_fill=["MAF"],verbose=False)
         is_filpped = (sumstats["MAF"] < sumstats[eaf]) & (sumstats[eaf] > 0.5)& (sumstats["MAF"] < 0.5)
-        if verbose: log.write(" -Flipping {} variants...".format(sum(is_filpped)))
+        log.write(" -Flipping {} variants...".format(sum(is_filpped)), verbose=verbose)
         sumstats.loc[is_filpped, beta] = -sumstats.loc[is_filpped, beta]
     
     #configure n #################################################################################################################
@@ -185,8 +183,7 @@ def plottrumpet(mysumstats,
             n = sumstats["N"].median() 
         elif n == "mean":
             n = sumstats["N"].mean() 
-        if verbose:
-            log.write(" -N for power calculation: {}".format(n))
+        log.write(" -N for power calculation: {}".format(n), verbose=verbose)
 
     #configure beta and maf range ###################################################################################################
     if maf_range is None:
@@ -307,6 +304,7 @@ def plottrumpet(mysumstats,
     
     if ylim is not None:
         ax.set_ylim(ylim)
+
     if yticks is not None:
         ax.set_yticks(yticks, yticklabels)
 
@@ -337,8 +335,8 @@ def plottrumpet(mysumstats,
             texts_d=[]
 
             if len(variants_toanno)>0:
-                
-                maxy = max(variants_toanno[beta].abs().max(),1.5)
+                maxy = variants_toanno[beta].abs().max()
+                #maxy = max(variants_toanno[beta].abs().max(),1.5)
                 variants_toanno["ADJUSTED_i"] = np.nan 
                 y_span = 0.5
                 
@@ -348,6 +346,9 @@ def plottrumpet(mysumstats,
                     variants_toanno = variants_toanno.sort_values(by=maf, key= np.abs, ascending = True)
                 
                 if anno_style == "expand":
+
+                    min_factor=None
+                    
                     if len(variants_toanno.loc[variants_toanno[beta]>0, "ADJUSTED_i"])>1:
                         variants_toanno.loc[variants_toanno[beta]>0, "ADJUSTED_i"] = adjust_text_position(variants_toanno.loc[variants_toanno[beta]>0,maf].values.copy(), 
                                                                                 y_span, 
@@ -355,7 +356,7 @@ def plottrumpet(mysumstats,
                                                                                 max_iter=anno_max_iter,
                                                                                 log=log,
                                                                                 amode=xscale,
-                                                                                verbose=verbose)
+                                                                                verbose=verbose,min_factor=min_factor)
 
                     if len(variants_toanno.loc[variants_toanno[beta]<0, "ADJUSTED_i"])>1:
                         variants_toanno.loc[variants_toanno[beta]<0, "ADJUSTED_i"] = adjust_text_position(variants_toanno.loc[variants_toanno[beta]<0,maf].values.copy(), 
@@ -364,10 +365,12 @@ def plottrumpet(mysumstats,
                                                                 max_iter=anno_max_iter,
                                                                 log=log,
                                                                 amode=xscale,
-                                                                verbose=verbose)
+                                                                verbose=verbose,min_factor=min_factor)
 
                 
                 for variants_toanno_half in [variants_toanno.loc[variants_toanno[beta]<0,:], variants_toanno.loc[variants_toanno[beta]>0,:]]:
+                    if len(variants_toanno_half)<1:
+                        continue
                     last_pos = min(variants_toanno_half[maf])/2
                     for index, row in variants_toanno_half.iterrows():
                         
@@ -395,6 +398,7 @@ def plottrumpet(mysumstats,
                         
                         if anno_style=="tight":
                             texts_d.append(ax.text(row[maf], row[beta], row[anno]))
+
                 if anno_style=="tight":
                     adjust_text(texts_d, 
                                 autoalign =True,
@@ -404,13 +408,15 @@ def plottrumpet(mysumstats,
                                 expand_points=(0.5,0.5),
                                 force_objects=(0.5,0.5), 
                                 ax=ax)
+    
+
     ############  Annotation ##################################################################################################
     if mode=="q":
         save_figure(fig, save, keyword="trumpet_q",save_args=save_args, log=log, verbose=verbose)
     elif mode=="b":
         save_figure(fig, save, keyword="trumpet_b",save_args=save_args, log=log, verbose=verbose)
 
-    if verbose: log.write("Finished creating trumpet plot!")
+    log.write("Finished creating trumpet plot!", verbose=verbose)
     return fig
 
 ####################################################################
@@ -460,12 +466,11 @@ def plot_power( ns=1000,
             xticklabels = xticks            
 
     #Checking columns#################################################################################################################
-    if verbose: log.write("Start to create trumpet plot...")
+    log.write("Start to create trumpet plot...", verbose=verbose)
     
     if mode=="b":
         if ncases is None or ncontrols is None:
-            if verbose:
-                log.write(" -No scase or scontrol. Skipping...")
+            log.write(" -No scase or scontrol. Skipping...", verbose=verbose)
             return None
         
     #configure beta and maf range ###################################################################################################
@@ -619,7 +624,7 @@ def plot_power( ns=1000,
     elif mode=="b":
         save_figure(fig, save, keyword="power_b",save_args=save_args, log=log, verbose=verbose)
 
-    if verbose: log.write("Finished creating trumpet plot!")
+    log.write("Finished creating trumpet plot!", verbose=verbose)
     return fig
 
 
@@ -658,29 +663,29 @@ def plot_power_x(
                 log=Log()):
     
     #Checking columns#################################################################################################################
-    if verbose: log.write("Start to create power plot...")
+    log.write("Start to create power plot...", verbose=verbose)
     matplotlib.rc('font', family=font_family)
-    if verbose:
-        log.write(" -Settings:")
-        log.write("  -Mode: {}".format(mode))
-        if mode == "q" :
-            log.write("  -X axis: {}".format(x))
-            if x!="N":
-                log.write("  -N: {}".format(ns))
-            if x!="MAF":
-                log.write("  -MAF: {}".format(mafs))
-        if mode == "b" :
-            log.write("  -X axis: {}".format(x))
-            if x!="N_CASE":
-                log.write("  -N_CASE: {}".format(ncases))
-            if x!="N_CASE":
-                log.write("  -N_CONTROL: {}".format(ncontrols))
-            if x!="PREVALENCE":
-                log.write("  -PREVALENCE: {}".format(prevalences))
-        if x!="BETA":
-            log.write("  -BETA: {}".format(betas))
-
-        log.write(" -Significance level: {}".format(sig_levels))
+
+    log.write(" -Settings:", verbose=verbose)
+    log.write("  -Mode: {}".format(mode), verbose=verbose)
+    if mode == "q" :
+        log.write("  -X axis: {}".format(x), verbose=verbose)
+        if x!="N":
+            log.write("  -N: {}".format(ns), verbose=verbose)
+        if x!="MAF":
+            log.write("  -MAF: {}".format(mafs), verbose=verbose)
+    if mode == "b" :
+        log.write("  -X axis: {}".format(x), verbose=verbose)
+        if x!="N_CASE":
+            log.write("  -N_CASE: {}".format(ncases), verbose=verbose)
+        if x!="N_CASE":
+            log.write("  -N_CONTROL: {}".format(ncontrols), verbose=verbose)
+        if x!="PREVALENCE":
+            log.write("  -PREVALENCE: {}".format(prevalences), verbose=verbose)
+    if x!="BETA":
+        log.write("  -BETA: {}".format(betas), verbose=verbose)
+
+    log.write(" -Significance level: {}".format(sig_levels), verbose=verbose)
 
     if x is None:
         if mode=="b":
@@ -698,8 +703,7 @@ def plot_power_x(
     
     if mode=="b":
         if ncases is None or ncontrols is None:
-            if verbose:
-                log.write(" -No scase or scontrol. Skipping...")
+            log.write(" -No scase or scontrol. Skipping...", verbose=verbose)
             return None
 
     #configure beta and maf range ###################################################################################################
@@ -914,5 +918,5 @@ def plot_power_x(
     elif mode=="b":
         save_figure(fig, save, keyword="power_xb",save_args=save_args, log=log, verbose=verbose)
 
-    if verbose: log.write("Finished creating power plot!")
+    log.write("Finished creating power plot!", verbose=verbose)
     return fig