gwaslab 3.5.0__tar.gz → 3.5.2__tar.gz

{gwaslab-3.5.0/src/gwaslab.egg-info → gwaslab-3.5.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gwaslab
-Version: 3.5.0
+Version: 3.5.2
 Summary: A collection of handy tools for GWAS SumStats
 Author-email: Yunye <yunye@gwaslab.com>
 Project-URL: Homepage, https://cloufield.github.io/gwaslab/
@@ -20,7 +20,7 @@ Requires-Dist: scipy>=1.12
 Requires-Dist: pySAM==0.22.1
 Requires-Dist: Biopython>=1.79
 Requires-Dist: adjustText<=0.8,>=0.7.3
-Requires-Dist: liftover>=1.1.13
+Requires-Dist: liftover<=1.3.1,>=1.1.13
 Requires-Dist: scikit-allel>=1.3.5
 Requires-Dist: pyensembl==2.2.3
 Requires-Dist: gtfparse==1.3.0
@@ -51,7 +51,7 @@ Warning: Known issues of GWASLab are summarized in [https://cloufield.github.io/
 ### install via pip
 
 ```
-pip install gwaslab==3.4.46
+pip install gwaslab==3.5.0
 ```
 
 ```python

{gwaslab-3.5.0 → gwaslab-3.5.2}/README.md

@@ -23,7 +23,7 @@ Warning: Known issues of GWASLab are summarized in [https://cloufield.github.io/
 ### install via pip
 
 ```
-pip install gwaslab==3.4.46
+pip install gwaslab==3.5.0
 ```
 
 ```python

{gwaslab-3.5.0 → gwaslab-3.5.2}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "gwaslab"
-version = "3.5.0"
+version = "3.5.2"
 authors = [
   { name="Yunye", email="yunye@gwaslab.com" },
 ]
@@ -24,7 +24,7 @@ dependencies = [
     "pySAM==0.22.1",
     "Biopython>=1.79",
     "adjustText>=0.7.3, <=0.8",
-    "liftover>=1.1.13",
+    "liftover>=1.1.13, <=1.3.1",
     "scikit-allel>=1.3.5",
     "pyensembl==2.2.3",
     "gtfparse==1.3.0",

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/__init__.py

@@ -46,3 +46,4 @@ from gwaslab.util_ex_process_h5 import process_vcf_to_hfd5
 from gwaslab.util_ex_run_susie import _run_susie_rss as run_susie_rss
 from gwaslab.io_read_tabular import _read_tabular as read_tabular
 from gwaslab.util_in_meta import meta_analyze
+from gwaslab.viz_plot_scatter_with_reg import scatter

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/bd_get_hapmap3.py

@@ -47,8 +47,10 @@ def gethapmap3(sumstats,rsid="rsID",chrom="CHR", pos="POS", ea="EA", nea="NEA",b
     #rsid    A1      A2      #CHROM  POS
     #rs3094315       G       A       1       752566
     
-    if rsid in sumstats.columns and how=="inner":
+    if rsid in sumstats.columns:
+        log.write(" -rsID will be used for matching...", verbose=verbose)
         output = sumstats.loc[sumstats[rsid].isin(hapmap3_ref["rsid"].values),:].copy()
+        log.write(" -Raw input contains "+str(len(output))+" Hapmap3 variants based on rsID...", verbose=verbose)
         return output
     
     elif chrom in sumstats.columns and pos in sumstats.columns:

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/g_Sumstats.py

@@ -389,6 +389,12 @@ class Sumstats():
             self.data =flipallelestats(self.data,log=self.log,**flipallelestats_args)
             
             gc.collect()
+            
+        if (ref_seq is not None or ref_infer is not None) and (ref_rsid_tsv is not None or ref_rsid_vcf is not None):
+
+            self.data = fixID(self.data, log=self.log, **{"fixid":True, "fixsep":True, "overwrite":True})
+
+            gc.collect()
         
         #####################################################
         if ref_rsid_tsv is not None:
@@ -790,28 +796,28 @@ class Sumstats():
     
     
 ## LDSC ##############################################################################################
-    def estimate_h2_by_ldsc(self, build=None, verbose=True, match_allele=True, **kwargs):
+    def estimate_h2_by_ldsc(self, build=None, verbose=True, match_allele=True, how="right", **kwargs):
         if build is None:
             build = self.meta["gwaslab"]["genome_build"]
-        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=True, how="right" )
+        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=match_allele, how=how )
         self.ldsc_h2, self.ldsc_h2_results = _estimate_h2_by_ldsc(insumstats=insumstats, log=self.log, verbose=verbose, **kwargs)
     
-    def estimate_rg_by_ldsc(self, build=None, verbose=True, match_allele=True, **kwargs):
+    def estimate_rg_by_ldsc(self, build=None, verbose=True, match_allele=True, how="right",**kwargs):
         if build is None:
             build = self.meta["gwaslab"]["genome_build"]
-        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=True, how="right" )
+        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=match_allele, how=how )
         self.ldsc_rg = _estimate_rg_by_ldsc(insumstats=insumstats, log=self.log, verbose=verbose, **kwargs)
 
-    def estimate_h2_cts_by_ldsc(self, build=None, verbose=True, match_allele=True, **kwargs):
+    def estimate_h2_cts_by_ldsc(self, build=None, verbose=True, match_allele=True, how="right",**kwargs):
         if build is None:
             build = self.meta["gwaslab"]["genome_build"]
-        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=True, how="right" )
+        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=match_allele, how=how )
         self.ldsc_h2_cts  = _estimate_h2_cts_by_ldsc(insumstats=insumstats, log=self.log, verbose=verbose, **kwargs)
     
-    def estimate_partitioned_h2_by_ldsc(self, build=None, verbose=True, match_allele=True, **kwargs):
+    def estimate_partitioned_h2_by_ldsc(self, build=None, verbose=True, match_allele=True, how="right",**kwargs):
         if build is None:
             build = self.meta["gwaslab"]["genome_build"]
-        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=True, how="right" )
+        insumstats = gethapmap3(self.data.copy(), build=build, verbose=verbose , match_allele=match_allele, how=how )
         self.ldsc_partitioned_h2_summary, self.ldsc_partitioned_h2_results  = _estimate_partitioned_h2_by_ldsc(insumstats=insumstats, log=self.log, verbose=verbose, **kwargs)
 # external ################################################################################################
     
@@ -833,4 +839,4 @@ class Sumstats():
     def to_format(self, path, build=None, verbose=True, **kwargs):
         if build is None:
             build = self.meta["gwaslab"]["genome_build"]
-        _to_format(self.data, path, log=self.log, verbose=verbose, meta=self.meta, build=build, **kwargs)
+        _to_format(self.data, path, log=self.log, verbose=verbose, meta=self.meta, build=build, **kwargs)

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/g_version.py

@@ -15,8 +15,8 @@ def _get_version():
 def gwaslab_info():
     # version meta information
     dic={
-   "version":"3.5.0",
-   "release_date":"20241029"
+   "version":"3.5.2",
+   "release_date":"20241203"
     }
     return dic   
 

gwaslab-3.5.2/src/gwaslab/io_process_args.py

@@ -0,0 +1,28 @@
+import copy
+
+def _list_func_args(func):
+    return func.__code__.co_varnames
+
+def _extract_kwargs(prefix:str, default:dict, kwargs:dict) -> dict:
+    # prefix: keyword
+    # default: default dict
+    # kwargs: all local kwargs  + args + kwargs
+
+    extracted = []
+    for key,value in kwargs.items():
+        if key=="kwargs" or key=="args":
+            for key_nested,value_nested in kwargs[key].items():
+                if prefix in key_nested and "arg" in key_nested:
+                    extracted.append(value_nested)
+        else:
+            if prefix in key and "arg" in key:
+                extracted.append(value)
+    merged_arg = _merge_and_sync_dic(extracted, default)
+    return merged_arg
+
+def _merge_and_sync_dic(list_of_dics:list, default:dict) -> dict:
+    temp = copy.copy(default)
+    for dic in list_of_dics:
+        if isinstance(dic, dict):
+            temp.update(dic)
+    return temp

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/qc_fix_sumstats.py

@@ -1579,14 +1579,14 @@ def liftover_variant(sumstats,
     
     try:
         if chain is None:
-            converter = get_lifter(from_build,to_build,one_based=True)
+            converter = get_lifter("hg{}".format(from_build),"hg{}".format(to_build),one_based=True)
         else:
-            converter = ChainFile(chain, one_based=True)
+            converter = ChainFile(chain,target="",query="", one_based=True)
     except:
         if chain is None:
-            converter = get_lifter(from_build,to_build)
+            converter = get_lifter("hg{}".format(from_build),"hg{}".format(to_build))
         else:
-            converter = ChainFile(chain)
+            converter = ChainFile(chain, target="",query="")
 
     dic= get_number_to_chr(in_chr=False,xymt=["X","Y","M"])
     dic2= get_chr_to_number(out_chr=False)
@@ -1627,11 +1627,15 @@ def parallelizeliftovervariant(sumstats,n_cores=1,chrom="CHR", pos="POS", from_b
     else:
         try:
             chain = get_chain(from_build=from_build, to_build=to_build)
-            log.write(" -Creating converter using ChainFile: {}".format(chain), verbose=verbose)
+            if chain is None or chain==False:
+                raise ValueError("")
+            log.write(" -Creating converter using provided ChainFile: {}".format(chain), verbose=verbose)
         except:
             chain = None
             lifter_from_build=from_build
             lifter_to_build=to_build
+            log.write(" -Try creating converter using liftover package", verbose=verbose)
+
     log.write(" -Creating converter : {} -> {}".format(lifter_from_build, lifter_to_build), verbose=verbose)
     # valid chr and pos
     pattern = r"\w\w\w0\w\w\w"  

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/viz_aux_annotate_plot.py

@@ -38,6 +38,7 @@ def annotate_single(
     region,
     region_anno_bbox_args,
     skip,
+    arrow_kwargs=None,
     anno_height=1,
     amode="int",
     snpid="SNPID",
@@ -48,6 +49,9 @@ def annotate_single(
     _invert=False,
     log=Log()
 ):
+    if arrow_kwargs is None:
+        arrow_kwargs=dict()
+        
     if anno and (to_annotate.empty is not True):
         #initiate a list for text and a starting position
         text = []
@@ -223,6 +227,8 @@ def annotate_single(
                 else:
                     arrowargs=dict(arrowstyle='-|>', color='grey', shrinkA=10, linewidth=0.1, relpos=(1,0.5))
             ################################################################################################################################
+            for key,value in arrow_kwargs.items():
+                arrowargs[key]=value
 
             anno_to_adjust = ax1.annotate(annotation_text,
                         xy=xy,
@@ -293,6 +299,7 @@ def annotate_pair(
     pos="POS",
     repel_force=0.02,
     verbose=True,
+    arrow_kwargs=None,
     log=Log()
 ):
     if anno is not None:
@@ -474,6 +481,7 @@ def annotate_pair(
                             arrowargs=dict(arrowstyle='-|>', color='grey', shrinkA=10, linewidth=0.1, relpos=(0,0.5))
                         else:
                             arrowargs=dict(arrowstyle='-|>', color='grey', shrinkA=10, linewidth=0.1, relpos=(1,0.5))
+
                     
                     anno_to_adjust = ax.annotate(annotation_text,
                             xy=xy,

gwaslab-3.5.2/src/gwaslab/viz_aux_property.py

@@ -0,0 +1,2 @@
+import matplotlib.pyplot as plt
+

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/viz_aux_quickfix.py

@@ -1,6 +1,8 @@
 import pandas as pd
 import numpy as np
 from gwaslab.g_Log import Log
+from matplotlib import ticker
+import matplotlib.pyplot as plt
 from gwaslab.bd_common_data import get_chr_to_number
 from gwaslab.bd_common_data import get_number_to_chr
 from math import ceil
@@ -350,7 +352,7 @@ def _cut(series, mode,cutfactor,cut,skip, ylabels, cut_log, verbose, lines_to_pl
                 #sumstats.loc[sumstats["scaled_P"]>cut,"scaled_P"] = (sumstats.loc[sumstats["scaled_P"]>cut,"scaled_P"]-cut)/cutfactor +  cut
                 
                 maxy = (maxticker-cut)/cutfactor + cut
-    
+
     return series, maxy, maxticker, cut, cutfactor,ylabels,lines_to_plot
 
 #def _cut_line(level, mode,cutfactor,cut,skip, ylabels, cut_log, verbose, log):
@@ -379,10 +381,9 @@ def _set_yticklabels(cut,
     log.write(" -Processing Y tick lables...",verbose=verbose)
     # if no cut
     if cut == 0: 
-            ax1.set_ylim(skip, ceil(maxy*1.2) )
-
+            ax1.set_ylim((skip, ceil(maxy*1.2)) )
     # if cut     
-    if cut:
+    if cut!=0:
         # add cut line
         
         cutline = ax1.axhline(y=cut, linewidth = sc_linewidth,linestyle="--",color=cut_line_color,zorder=1)
@@ -432,14 +433,15 @@ def _set_yticklabels(cut,
         else:
             ax1.set_yticks(ticks1+ticks2)
             ax1.set_yticklabels(tickslabel1+tickslabel2,fontsize=fontsize,family=font_family)
-        ax1.set_ylim(bottom = skip)
     
     if ylabels is not None:  
         ax1.set_yticks(ylabels_converted)
         ax1.set_yticklabels(ylabels,fontsize=fontsize,family=font_family)
     
+    ylim_top = ax1.get_ylim()[1]
+    ax1.set_ybound(lower=skip,upper=ylim_top)
     ax1.tick_params(axis='y', labelsize=fontsize)
-    
+
     return ax1
 
 def _jagged_y(cut,skip,ax1,mode,mqqratio,jagged_len,jagged_wid, log=Log(), verbose=True):

{gwaslab-3.5.0 → gwaslab-3.5.2}/src/gwaslab/viz_aux_save_figure.py

@@ -51,7 +51,8 @@ def get_default_path(keyword,fmt="png"):
                         "miami":"miami",
                         "esc":"effect_size_comparision",
                         "afc":"allele_frequency_comparision",
-                        "gwheatmap":"genome_wide_heatmap"
+                        "gwheatmap":"genome_wide_heatmap",
+                        "scatter":"scatter"
                         }
     prefix = path_dictionary[keyword]
     count = 1