gwaslab 3.5.4__py3-none-any.whl → 3.5.6__py3-none-any.whl

gwaslab/viz_plot_effect.py

@@ -0,0 +1,283 @@
+import pandas as pd
+import matplotlib.pyplot as plt
+import matplotlib.ticker as ticker
+import matplotlib.patches as patches
+import seaborn as sns
+import numpy as np
+import scipy as sp
+import copy
+from math import ceil
+from shutil import which
+from pyensembl import EnsemblRelease
+from allel import GenotypeArray
+from allel import read_vcf
+from allel import rogers_huff_r_between
+import matplotlib as mpl
+from scipy import stats
+from mpl_toolkits.axes_grid1.inset_locator import inset_axes
+from mpl_toolkits.axes_grid1.inset_locator import mark_inset
+from matplotlib.ticker import MaxNLocator
+import gc as garbage_collect
+from adjustText import adjust_text
+from gwaslab.viz_aux_reposition_text import adjust_text_position
+from gwaslab.viz_aux_annotate_plot import annotate_single
+from gwaslab.viz_plot_qqplot import _plot_qq
+from gwaslab.hm_harmonize_sumstats import auto_check_vcf_chr_dict
+from gwaslab.viz_plot_regional2 import _plot_regional
+from gwaslab.viz_plot_regional2 import process_vcf
+from gwaslab.viz_plot_regional2 import _get_lead_id
+from gwaslab.viz_aux_quickfix import _get_largenumber
+from gwaslab.viz_aux_quickfix import _quick_fix_p_value
+from gwaslab.viz_aux_quickfix import _quick_fix_pos
+from gwaslab.viz_aux_quickfix import _quick_fix_chr
+from gwaslab.viz_aux_quickfix import _quick_fix_eaf
+from gwaslab.viz_aux_quickfix import _quick_fix_mlog10p
+from gwaslab.viz_aux_quickfix import _quick_add_tchrpos
+from gwaslab.viz_aux_quickfix import _quick_merge_sumstats
+from gwaslab.viz_aux_quickfix import _quick_assign_i
+from gwaslab.viz_aux_quickfix import _quick_assign_i_with_rank
+from gwaslab.viz_aux_quickfix import _quick_extract_snp_in_region
+from gwaslab.viz_aux_quickfix import _quick_assign_highlight_hue_pair
+from gwaslab.viz_aux_quickfix import _quick_assign_marker_relative_size
+from gwaslab.viz_aux_quickfix import _cut
+from gwaslab.viz_aux_quickfix import _set_yticklabels
+from gwaslab.viz_aux_quickfix import _jagged_y
+from gwaslab.viz_aux_save_figure import save_figure
+from gwaslab.g_Log import Log
+from gwaslab.util_in_calculate_gc import lambdaGC
+from gwaslab.util_in_get_sig import getsig
+from gwaslab.util_in_get_sig import annogene
+from gwaslab.bd_common_data import get_chr_to_number
+from gwaslab.bd_common_data import get_number_to_chr
+from gwaslab.bd_common_data import get_recombination_rate
+from gwaslab.bd_common_data import get_gtf
+from gwaslab.util_in_filter_value import _filter_region
+from gwaslab.g_version import _get_version
+from matplotlib.colors import ListedColormap
+from matplotlib.colors import LinearSegmentedColormap
+from matplotlib.colors import to_hex
+from gwaslab.io_process_args import _extract_kwargs
+
+def _plot_effect(to_plot, 
+                 y=None, 
+                 y_sort=None, 
+                 group=None, 
+                 x="BETA", 
+                 se="SE", 
+                 eaf="EAF", 
+                 snpr2="SNPR2", 
+                 ylabel="Variant",
+                 eaf_panel=True, 
+                 snpvar_panel=True, 
+                 rename_dic=None, 
+                 err_args=None,
+                 font_args=None,
+                 save=None,
+                 title=None,
+                 save_args=None,
+                 eaf_args=None,
+                 snpr2_args=None,
+                 fig_args=None,
+                 scatter_args=None,
+                 effect_label=None,
+                 eaf_label=None,
+                 snpr2_label=None,
+                 log=Log(),
+                 verbose=True,
+                 legend_mode=1,
+                 ncol=2,
+                 size=None,
+                 hue=None,
+                 style=None,
+                 **args):
+
+    if err_args is None:
+        err_args={"ecolor":"#cccccc",
+                  "linewidth":0,
+                  "zorder":90,
+                  "elinewidth":1}
+    if eaf_args is None:
+        eaf_args={"color":"#74BAD3"}
+    if snpr2_args is None:
+        snpr2_args={"color":"#74BAD3"}
+    if font_args is None:
+        font_args={'fontsize':12,'family':'sans','fontname':'Arial'}
+    if fig_args is None:
+        fig_args={"figsize":(8,8),"dpi":300}
+    if scatter_args is None:
+        scatter_args={"s":20}
+    
+
+    legend_titles=[]
+    if hue is not None:
+        args["hue"] = hue
+        legend_titles.append(hue)
+
+    if size is not None:
+        args["size"] = size
+        legend_titles.append(size)
+
+    if style is not None:
+        args["style"] = style
+        legend_titles.append(style)
+
+        
+    save_kwargs =      _extract_kwargs("save", save_args, locals())
+    err_kwargs =       _extract_kwargs("err", err_args, locals())
+    scatter_kwargs =   _extract_kwargs("scatter", scatter_args, locals())
+    font_kwargs =      _extract_kwargs("font",font_args, locals())
+    
+    def concat_cols(cols):
+        string = "-".join(map(str,cols))
+        return string
+    
+    y_name = "-".join(y)
+    
+    to_plot[y_name] = to_plot[y].apply(lambda x: concat_cols(x), axis=1)
+    
+
+    # sort y 
+    if y_sort is None:
+        y_sort = ["CHR","POS","STUDY"]
+    
+    to_plot = to_plot.sort_values(by=y_sort)
+
+    if group is None:
+        group = ["CHR","POS"]
+        to_plot = to_plot.sort_values(by=group)
+    # Assign group IDs based on the sorted 'score'
+        
+    to_plot['_VAR_GROUP'] = to_plot.groupby(group).ngroup() + 1
+
+    to_plot["_VAR_INDEX"] = range(len(to_plot))
+    to_plot["_VAR_INDEX"]= to_plot["_VAR_INDEX"] + to_plot['_VAR_GROUP']
+
+    y="_VAR_INDEX"
+
+    if rename_dic is None:
+        rename_dic = {
+            "BETA":"Per-allele effect size",
+            "STUDY":"Study"
+                      }
+    ncols=1
+    if eaf_panel:
+        ncols+=1
+    if snpvar_panel:
+        ncols+=1
+
+    if ncols==1:
+        fig,ax1 = plt.subplots(ncols=ncols, **fig_args)
+    elif ncols==2:
+        if eaf_panel==True:
+            fig,axes = plt.subplots(ncols=ncols, dpi=400,sharey=True)
+            ax1=axes[0]
+            ax2=axes[1]
+        else:
+            fig,axes = plt.subplots(ncols=ncols, dpi=400,sharey=True)
+            ax1=axes[0]
+            ax3=axes[1]
+    else:
+        fig,axes = plt.subplots(ncols=ncols, dpi=400,sharey=True)
+        ax1=axes[0]
+        ax2=axes[1]
+        ax3=axes[2]
+
+    sns.scatterplot(data=to_plot, x=x, y=y, ax=ax1, zorder=100, **args)
+
+    ax1.errorbar(y=to_plot[y], x=to_plot[x], xerr=to_plot[se], 
+                  **err_kwargs)
+    
+    ax1.axvline(x=0,linestyle="dashed",c="grey")
+    ax1.set_yticks(to_plot[y], labels = to_plot[y_name])
+    ax1.set_ylabel(ylabel) 
+
+    if title is not None:
+        ax1.set_title(title)
+
+    if eaf_panel==True:
+        ax2.barh(y=to_plot[y], width=to_plot[eaf], zorder=100, **eaf_args)
+        ax2.set_xlabel(eaf)
+
+    if snpvar_panel==True:
+        ax3.barh(y=to_plot[y], width=to_plot[snpr2], zorder=100,**snpr2_args)
+        ax3.set_xlabel(snpr2)
+    
+    #try:
+    if legend_mode==1:
+        #if ncols==1:
+        sns.move_legend(
+            ax1, "upper left",
+            bbox_to_anchor=(1, 1), title=None, frameon=False, bbox_transform = axes[-1].transAxes
+            )
+            #else:
+##
+            #    sns.move_legend(
+            #        ax1, "lower left",
+            #        bbox_to_anchor=(0, ncols), title=None, frameon=False,
+            #    )
+        #elif legend_mode==2:
+        #    sns.move_legend(
+        #        ax1, "lower center",
+        #        bbox_to_anchor=(0, 1), ncol=ncol, title=None, frameon=False,
+        #        )
+    #except:
+    #    pass
+
+
+    #handles, labels = ax1.get_legend_handles_labels()
+    #if len(labels)>0:
+    #    #new_labels = []
+    #    #ncol = len(labels)
+    #    max_col=0
+    #    new_labels=[]
+    #    new_labels_i = []
+    #    previous_i = 0
+    #    max_string_len=0
+    #    for i in range(len(labels)):
+    #        if len(labels[i]) > max_string_len:
+    #            max_string_len = len(labels[i])
+    #        if labels[i] in legend_titles:
+    #            new_labels_i.append(i)
+    #            col_number = i - previous_i
+    #            if col_number > max_col:
+    #                max_col = col_number
+    #            previous_i = i
+    #    for i in labels:
+    #        new_labels.append(str(i).ljust(max_string_len))
+    #    print(new_labels)
+    #    new_labels_i.append(len(labels))
+#
+    #    legend_rows = []
+    #    #new_labels_i[index+1] - i
+    #    for index, i in enumerate(new_labels_i):
+    #        if index<len(new_labels_i)-1:
+    #            legend_row = ax1.legend(labels = new_labels[i:new_labels_i[index+1]],  
+    #                                    handles= handles[i:new_labels_i[index+1]],
+    #                                    loc="lower left", 
+    #                                    bbox_to_anchor=(-0.2, 1.02 + 0.05*index), 
+    #                                    ncol=max_col, 
+    #                                    scatterpoints=1, 
+    #                                    title=None, 
+    #                                    borderpad=0,
+    #                                    handletextpad=0.1,
+    #                                    handlelength=0.7,
+    #                                    borderaxespad =0,
+    #                                    alignment = "left",
+    #                                    fontsize=8,
+    #                                    frameon=False)
+    #            legend_rows.append(legend_row)
+    #    for legend_row in legend_rows[:-1]:
+    #        ax1.add_artist(legend_row)
+
+
+
+    if effect_label is not None:
+        ax1.set_xlabel(effect_label)
+    if eaf_label is not None:
+        ax2.set_xlabel(eaf_label)
+    if snpr2_label is not None:
+        ax3.set_xlabel(snpr2_label)
+
+    save_figure(fig, save, keyword="forest",save_args=save_kwargs, log=log, verbose=verbose)   
+    
+    return fig

gwaslab/viz_plot_miamiplot2.py

@@ -249,7 +249,7 @@ def plot_miami2(
     #####################################################################################################################
     ##plotting
     if figax is None:
-        fig_args["figsize"] = (15,10)
+        #fig_args["figsize"] = (15,10)
         fig, (ax1, ax5) = plt.subplots(2, 1, gridspec_kw={'height_ratios': [1, 1]},**fig_args)
         plt.subplots_adjust(hspace=region_hspace)    
     else:

gwaslab/viz_plot_mqqplot.py

@@ -51,6 +51,7 @@ from gwaslab.bd_common_data import get_chr_to_number
 from gwaslab.bd_common_data import get_number_to_chr
 from gwaslab.bd_common_data import get_recombination_rate
 from gwaslab.bd_common_data import get_gtf
+from gwaslab.util_in_filter_value import _filter_region
 from gwaslab.g_version import _get_version
 from matplotlib.colors import ListedColormap
 from matplotlib.colors import LinearSegmentedColormap
@@ -109,9 +110,17 @@ def mqqplot(insumstats,
           region_anno_bbox_args = None,
           region_marker_shapes=None,
           region_legend_marker=True,
+          region_ref_alias = None,
           cbar_title='LD $r^{2}$ with variant',
           cbar_fontsize = None,
+          cbar_scale=True,
           cbar_font_family = None,
+          cbar_bbox_to_anchor = (0,0,1,1),
+          cbar_equal_aspect = True,
+          cbar_w_scale=1,
+          cbar_h_scale=1,
+          cbar_downward_offset =1.3, 
+          cbar_borderpad=None,
           track_n=4,
           track_n_offset=0,
           track_fontsize_ratio=0.95,
@@ -475,18 +484,20 @@ def mqqplot(insumstats,
         sumstats[chrom] = _quick_fix_chr(sumstats[chrom], chr_dict=chr_dict)
 
     ## r
+    
     if region is not None:
-        region_chr = region[0]
-        region_start = region[1]
-        region_end = region[2]
-        
-        log.write(" -Extract SNPs in region : chr{}:{}-{}...".format(region_chr, region[1], region[2]),verbose=verbose)
-        
-        in_region_snp = (sumstats[chrom]==region_chr) & (sumstats[pos]<region_end) & (sumstats[pos]>region_start)
-        
-        log.write(" -Extract SNPs in specified regions: "+str(sum(in_region_snp)),verbose=verbose)
-        sumstats = sumstats.loc[in_region_snp,:]
-        
+        sumstats = _filter_region(sumstats, region, log=log, verbose=verbose)
+    #    region_chr = region[0]
+    #    region_start = region[1]
+    #    region_end = region[2]
+    #    
+    #    log.write(" -Extract SNPs in region : chr{}:{}-{}...".format(region_chr, region[1], region[2]),verbose=verbose)
+    #    
+    #    in_region_snp = (sumstats[chrom]==region_chr) & (sumstats[pos]<region_end) & (sumstats[pos]>region_start)
+    #    
+    #    log.write(" -Extract SNPs in specified regions: "+str(sum(in_region_snp)),verbose=verbose)
+    #    sumstats = sumstats.loc[in_region_snp,:]
+    #    
         if len(sumstats)==0:
             log.warning("No valid data! Please check the input.")
             return None
@@ -810,6 +821,14 @@ def mqqplot(insumstats,
                                 fontsize=fontsize,
                                 build=build,
                                 chrom_df=chrom_df,
+                                cbar_scale=cbar_scale,
+                                cbar_fontsize=cbar_fontsize,
+                                cbar_bbox_to_anchor=cbar_bbox_to_anchor,
+                                cbar_w_scale=cbar_w_scale,
+                                cbar_h_scale=cbar_h_scale,
+                                cbar_equal_aspect=cbar_equal_aspect,
+                                cbar_downward_offset =cbar_downward_offset, 
+                                cbar_borderpad=cbar_borderpad,
                                 xtick_chr_dict=xtick_chr_dict,
                                 cut_line_color=cut_line_color,
                                 vcf_chr_dict =vcf_chr_dict,
@@ -825,6 +844,7 @@ def mqqplot(insumstats,
                                 region_step = region_step,
                                 region_ref = region_ref,
                                 region_ref_index_dic = region_ref_index_dic,
+                                region_ref_alias = region_ref_alias,
                                 region_grid = region_grid,
                                 region_grid_line = region_grid_line,
                                 region_lead_grid = region_lead_grid,

gwaslab/viz_plot_regional2.py

@@ -57,6 +57,7 @@ def _plot_regional(
     region_step = 21,
     region_ref=None,
     region_ref_index_dic = None,
+    region_ref_alias = None,
     #region_ref_second=None,
     region_grid = False,
     region_grid_line = {"linewidth": 2,"linestyle":"--"},
@@ -67,6 +68,14 @@ def _plot_regional(
     region_ld_threshold = [0.2,0.4,0.6,0.8],
     region_ld_colors = ["#E4E4E4","#020080","#86CEF9","#24FF02","#FDA400","#FF0000","#FF0000"],
     region_marker_shapes=None,
+    cbar_fontsize=None,
+    cbar_scale=False,
+    cbar_bbox_to_anchor=None,
+    cbar_w_scale=1,
+    cbar_h_scale=1,
+    cbar_downward_offset =1.3, 
+    cbar_borderpad=None,
+    cbar_equal_aspect=False,
     palette=None,
     region_recombination = True,
     region_protein_coding=True,
@@ -132,7 +141,16 @@ def _plot_regional(
                             region_ref=region_ref,
                             region_ld_threshold=region_ld_threshold, 
                             region_ref_index_dic=region_ref_index_dic,
+                            region_ref_alias=region_ref_alias, 
                             region_marker_shapes=region_marker_shapes,
+                            cbar_fontsize= cbar_fontsize,
+                            cbar_scale=cbar_scale,
+                            cbar_equal_aspect=cbar_equal_aspect,
+                            cbar_bbox_to_anchor=cbar_bbox_to_anchor,
+                            cbar_w_scale=cbar_w_scale,
+                            cbar_h_scale=cbar_h_scale,
+                            cbar_downward_offset =cbar_downward_offset, 
+                            cbar_borderpad=cbar_borderpad,
                             palette=palette,
                             region_legend_marker=region_legend_marker,
                             fig=fig)
@@ -154,6 +172,8 @@ def _plot_regional(
                                         build= build,
                                         rr_lim=rr_lim,
                                         rr_ylabel=rr_ylabel)
+    else:
+        ax4 = None
      
     ## regional plot : gene track ######################################################################
                 # calculate offset
@@ -345,25 +365,52 @@ def _add_region_title(region_title, ax1,region_title_args):
     ax1.text(0.015,0.97, region_title, transform=ax1.transAxes, va="top", ha="left", region_ref=None, **region_title_args )
     return ax1
 
-def _add_ld_legend(sumstats, ax1, region_ld_threshold, region_ref,region_ref_index_dic,region_marker_shapes,fig, region_legend_marker=True,palette =None, position=1):
+def _add_ld_legend(sumstats, ax1, region_ld_threshold, region_ref,region_ref_index_dic,region_marker_shapes,fig, region_legend_marker=True,
+                   cbar_fontsize= None,cbar_scale=False,cbar_equal_aspect=True,cbar_w_scale=1,cbar_h_scale=1,palette =None, 
+                   cbar_downward_offset =1.2, cbar_borderpad=None,
+                   cbar_bbox_to_anchor=(0, 0, 1, 1),region_ref_alias=None):
+
+    scale = 1
+    if cbar_scale:
+        base_fontsize = 9
+        scale = cbar_fontsize / base_fontsize
+        scale = max(1,scale)
+    else:
+        scale = 1
+    
+    width_raw= 11 * (scale)*cbar_w_scale
+    height_raw=(7 + 7 * len(region_ref))*(scale)*cbar_h_scale
+
+    width_pct = "{}%".format(width_raw)
+    height_pct = "{}%".format( height_raw)
+
+    total_y_pixels =(ax1.bbox.get_points()[1][1]-ax1.bbox.get_points()[0][1]) 
+    downwards_offset = cbar_fontsize / (total_y_pixels/ fig.dpi * 72) * cbar_downward_offset
+    bbox_to_anchor = (cbar_bbox_to_anchor[0],cbar_bbox_to_anchor[1]-downwards_offset,cbar_bbox_to_anchor[2],cbar_bbox_to_anchor[3])
     
-    width_pct = "11%"
-    height_pct = "{}%".format( 14 + 7 * len(region_ref))
+    if cbar_borderpad is None:
+        borderpad=0.5*(scale)
+    else:
+        borderpad=cbar_borderpad
+
     axins1 = inset_axes(ax1,
             width=width_pct,  # width = 50% of parent_bbox width
             height=height_pct,  # height : 5%
+            bbox_to_anchor=bbox_to_anchor,
+            bbox_transform=ax1.transAxes,
+            borderpad=borderpad,
             loc='upper right',
-            axes_kwargs={"frameon":True,"facecolor":"white","zorder":999999})
-    
+            axes_kwargs={"frameon":True,"facecolor":"white","zorder":999999,"anchor":"NE"})
+
     ld_ticks = [0]+region_ld_threshold+[1]
 
     for index, ld_threshold in enumerate(ld_ticks):
         for group_index in range(len(region_ref)):
             if index < len(ld_ticks)-1:            
                 x=ld_threshold
-                y=0.2*group_index
+                y=0.2*group_index 
                 width=0.2
-                height=ld_ticks[index+1]-ld_ticks[index]
+                height=(ld_ticks[index+1]-ld_ticks[index]) 
                 hex_color = palette[(region_ref_index_dic[region_ref[group_index]]+1)*100 + index+1] # consistent color
                 
                 a = Rectangle((x,y),width, height, fill = True, color = hex_color , linewidth = 2)
@@ -371,22 +418,46 @@ def _add_ld_legend(sumstats, ax1, region_ld_threshold, region_ref,region_ref_ind
                 axins1.add_patch(a)
     
     # y snpid
-    yticks_position = 0.1 + 0.2 *np.arange(0,len(region_ref))
-    axins1.set_yticks(yticks_position, ["{}".format(x) for x in region_ref])
-    axins1.set_ylim(0,0.2*len(region_ref))    
-    
+    if region_ref_alias is None:
+        region_ref_name = region_ref
+    else:
+        region_ref_name = [region_ref_alias[i] for i in region_ref]
+
+    yticks_position = (0.1 + 0.2 *np.arange(0,len(region_ref_name)))
+    axins1.set_yticks(yticks_position, ["{}".format(x) for x in region_ref_name])
+    axins1.set_ylim(0,0.2*len(region_ref_name))    
+    ymin, ymax=0,0.2*len(region_ref_name)
     # x ld thresholds
+    
     axins1.set_xticks(ticks=ld_ticks) 
     axins1.set_xticklabels([str(i) for i in ld_ticks]) 
-
     xmin, xmax = 0, 1
-    axins1.set_xlim(xmin,xmax)    
+    axins1.set_xlim(xmin,xmax)       
+
+    if cbar_equal_aspect==True:
+        axins1.set_aspect('equal', adjustable='box',anchor="NE")
 
     ############### ##############plot marker ############## ##############
     if region_legend_marker==True:
         for group_index, ref in enumerate(region_ref):
-            x= -0.1
-            y= 0.1 + 0.2 * group_index
+
+            data_to_point_y =((axins1.bbox.get_points()[1][1]-axins1.bbox.get_points()[0][1])*height_raw/(ymax -ymin))
+            data_to_point_x =((axins1.bbox.get_points()[1][0]-axins1.bbox.get_points()[0][0])*width_raw/(xmax -xmin))
+            y_to_x = data_to_point_y/data_to_point_x
+            x_to_y = 1/y_to_x
+            xyratio = min(y_to_x, x_to_y)
+            
+            marker_side_in_data = 0.075
+            if cbar_equal_aspect==True:
+                xyratio=1
+            
+            ## change markersize
+
+            if xyratio <1 :
+                x = 0 - (marker_side_in_data +0.03) * xyratio
+            else:
+                x = 0 - (marker_side_in_data +0.03)
+            y= (0.1 + 0.2 * group_index)
             
             if len(region_ref) <2:
                 # single-ref mode
@@ -398,16 +469,31 @@ def _add_ld_legend(sumstats, ax1, region_ld_threshold, region_ref,region_ref_ind
                 c =  palette[(region_ref_index_dic[region_ref[group_index]]+1)*100 + len(ld_ticks)-1]
             
             # ([x0,y0][x1,y1])
-            data_to_point =(axins1.bbox.get_points()[1][0]-axins1.bbox.get_points()[0][0]) / (xmax - xmin)
-            s =  (data_to_point * 0.15*0.11/(fig.dpi/72))**2 
+            #  y pixels / per data 1
+            
+            data_to_point_y =((axins1.bbox.get_points()[1][1]-axins1.bbox.get_points()[0][1])*height_raw/(ymax -ymin))
+            data_to_point_x =((axins1.bbox.get_points()[1][0]-axins1.bbox.get_points()[0][0])*width_raw/(xmax -xmin))
+            
+            if data_to_point_y < data_to_point_x:
+                length_raw = 1 #height_raw
+                data_to_point = data_to_point_y
+            else:
+                length_raw = 1 #width_raw
+                data_to_point = data_to_point_x
+
+            # pixels/data 1 -> font points/data 1  
+            #  (dpi / 72) = point_per_pixel
+            # y pixels / per data 1 / (dpi / 72)  -> y font points/data 1  
+            
+            font_points_per_data_1 = data_to_point/(fig.dpi/72)
+            s =  ((marker_side_in_data*2)* font_points_per_data_1 * length_raw/100 )**2     
             
             axins1.scatter(x, y, s=s, marker=marker,c=c, edgecolors="black", linewidths = 1,  clip_on=False, zorder=100)
-            axins1.tick_params(axis="y", pad=data_to_point* 0.11* 0.19/(fig.dpi/72))
-    
-    axins1.set_xlim(0,1)   
-    axins1.set_aspect('equal', adjustable='box')
-    #axins1.tick_params(axis="y", pad=data_to_point* 0.11* 0.19/(fig.dpi/72))
-    axins1.set_title('LD $r^{2}$ with variant',loc="center",y=-0.2)
+
+            pad = ((marker_side_in_data*2+0.02)* font_points_per_data_1 * length_raw/100)
+            tick_length=(abs(x)* font_points_per_data_1 * length_raw/100)
+            axins1.tick_params(axis="y", pad=pad-0.5*tick_length, length=tick_length) 
+
     cbar = axins1
     return ax1, cbar
 
@@ -479,12 +565,15 @@ def _plot_gene_track(
     stack_num_to_plot = max(taf[0],n_uniq_stack)
     ax3.set_ylim((-stack_num_to_plot*2-taf[1]*2,2+taf[1]*2))
     ax3.set_yticks([])
-    pixels_per_point = 72/fig.dpi
+    point_per_pixels = 72/fig.dpi
+    pixels_per_point = fig.dpi/72
+
     pixels_per_track = np.abs(ax3.transData.transform([0,0])[1] - ax3.transData.transform([0,1])[1])                   
     font_size_in_pixels= taf[2] * pixels_per_track
-    font_size_in_points =  font_size_in_pixels * pixels_per_point
-    linewidth_in_points=   pixels_per_track * pixels_per_point
+    font_size_in_points =  font_size_in_pixels * point_per_pixels
 
+    linewidth_in_points_per_track=   pixels_per_track * point_per_pixels
+    
     log.write(" -plotting gene track..", verbose=verbose)
     
     sig_gene_name = "Undefined"
@@ -517,8 +606,11 @@ def _plot_gene_track(
                 sig_gene_rights.append(gene_track_start_i+row["end"])
 
         # plot gene line
+        ## minimum width = 2 pixel 
+        gene_line_width = max(linewidth_in_points_per_track/10, 2/pixels_per_point)
+        
         ax3.plot((gene_track_start_i+row["start"],gene_track_start_i+row["end"]),
-                    (row["stack"]*2,row["stack"]*2),color=gene_color,linewidth=linewidth_in_points/10,solid_capstyle="butt")
+                    (row["stack"]*2,row["stack"]*2),color=gene_color,linewidth=gene_line_width,solid_capstyle="butt")
         
         # plot gene name
         if row["end"] >= region[2]:
@@ -549,9 +641,12 @@ def _plot_gene_track(
                 exon_color = region_lead_grid_line["color"]  
             else:
                 exon_color="#020080"
-        
+
+        ## minimum width = 8 pixel 
+        exon_line_width = max(linewidth_in_points_per_track * taf[3],  8/pixels_per_point) 
+
         ax3.plot((gene_track_start_i+row["start"],gene_track_start_i+row["end"]),
-                    (row["stack"]*2,row["stack"]*2),linewidth=linewidth_in_points*taf[3],color=exon_color,solid_capstyle="butt")
+                    (row["stack"]*2,row["stack"]*2),linewidth=exon_line_width,color=exon_color,solid_capstyle="butt")
 
     log.write(" -Finished plotting gene track..", verbose=verbose)
 
@@ -573,7 +668,6 @@ def process_vcf(sumstats,
                 region_ld_threshold, 
                 vcf_chr_dict,
                 tabix):
-
     log.write("Start to load reference genotype...", verbose=verbose)
     log.write(" -reference vcf path : "+ vcf_path, verbose=verbose)
 
@@ -631,7 +725,6 @@ def process_vcf(sumstats,
             # figure out lead variant
             lead_id = _get_lead_id(sumstats, region_ref_single, log, verbose)
         
-
         lead_series = None
         if lead_id is None:
             
@@ -685,6 +778,12 @@ def process_vcf(sumstats,
         else:
             log.write(" -Lead SNP not found in reference...", verbose=verbose)
             sumstats[rsq]=None
+            
+            # 
+            try:
+                sumstats.loc[lead_id,rsq]=1
+            except KeyError:
+                pass
         
         sumstats[rsq] = sumstats[rsq].astype("float")
         sumstats[ld_single] = 0
@@ -716,7 +815,7 @@ def process_vcf(sumstats,
     sumstats[final_ld_col]  = 0
     sumstats[final_shape_col] = 1
     sumstats[final_rsq_col] = 0.0
-    
+
     if len(region_ref)==1:
         if lead_id is not None:
             sumstats.loc[lead_id, final_shape_col] +=1 
@@ -729,6 +828,8 @@ def process_vcf(sumstats,
         sumstats.loc[a_ngt_b, final_ld_col] = 100 * (i+1) + sumstats.loc[a_ngt_b, ld_single]
         sumstats.loc[a_ngt_b, final_rsq_col] = sumstats.loc[a_ngt_b, current_rsq]
         sumstats.loc[a_ngt_b, final_shape_col] = i + 1
+    
+    sumstats = sumstats.dropna(subset=[pos,nea,ea])
     ####################################################################################################
     log.write("Finished loading reference genotype successfully!", verbose=verbose)
     return sumstats