gwaslab 3.4.49__tar.gz → 3.5.1__tar.gz

{gwaslab-3.4.49/src/gwaslab.egg-info → gwaslab-3.5.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gwaslab
-Version: 3.4.49
+Version: 3.5.1
 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-3.4.49 → gwaslab-3.5.1}/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-3.4.49 → gwaslab-3.5.1}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "gwaslab"
-version = "3.4.49"
+version = "3.5.1"
 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-3.4.49 → gwaslab-3.5.1}/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-3.4.49 → gwaslab-3.5.1}/src/gwaslab/g_Sumstats.py

@@ -81,6 +81,7 @@ from gwaslab.bd_get_hapmap3 import gethapmap3
 from gwaslab.util_abf_finemapping import abf_finemapping
 from gwaslab.util_abf_finemapping import make_cs
 import gc
+from gwaslab.viz_plot_phe_heatmap import _gwheatmap
 
 #20220309
 class Sumstats():
@@ -602,6 +603,11 @@ class Sumstats():
     def plot_daf(self, **kwargs):
         fig,outliers = plotdaf(self.data, **kwargs)
         return fig, outliers
+    
+    def plot_gwheatmap(self, **kwargs):
+        fig = _gwheatmap(self.data, **kwargs)
+        return fig
+    
     def plot_mqq(self, build=None, **kwargs):
 
         chrom="CHR"
@@ -705,7 +711,7 @@ class Sumstats():
         # return sumstats object    
         return output
     
-    def check_cis(self, **kwargs):
+    def check_cis(self, gls=False, **kwargs):
         if "SNPID" in self.data.columns:
             id_to_use = "SNPID"
         else:
@@ -717,7 +723,13 @@ class Sumstats():
                            p="P",
                            log=self.log,
                            **kwargs)
-        # return sumstats object    
+        
+        # return sumstats object   
+        if gls == True:
+            new_Sumstats_object = copy.deepcopy(self)
+            new_Sumstats_object.data = output
+            gc.collect()
+            return new_Sumstats_object
         return output
     
     def check_novel_set(self, **kwargs):
@@ -778,28 +790,28 @@ 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 ################################################################################################
     

{gwaslab-3.4.49 → gwaslab-3.5.1}/src/gwaslab/g_version.py

@@ -15,8 +15,8 @@ def _get_version():
 def gwaslab_info():
     # version meta information
     dic={
-   "version":"3.4.49",
-   "release_date":"20241019"
+   "version":"3.5.1",
+   "release_date":"20241120"
     }
     return dic   
 

gwaslab-3.5.1/src/gwaslab/io_process_args.py

@@ -0,0 +1,23 @@
+import copy
+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
+
+def _list_func_args(func):
+    return func.__code__.co_varnames
+
+def _extract_kwargs(prefix:str, default:dict, kwargs:dict) -> dict:
+    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

{gwaslab-3.4.49 → gwaslab-3.5.1}/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-3.4.49 → gwaslab-3.5.1}/src/gwaslab/util_in_get_sig.py

@@ -619,8 +619,10 @@ def _check_cis(insumstats,
     except:
         pass
 
-    allsig["CIS/TRANS"] = allsig.apply(lambda x: determine_if_cis(x, group_key,windowsizekb, reference_dict), axis=1)
-    
+    #allsig["CIS/TRANS"] = allsig.apply(lambda x: determine_if_cis(x, group_key,windowsizekb, reference_dict), axis=1)
+    cis_tuples = allsig.apply(lambda x: determine_if_cis2(x, group_key,windowsizekb, reference_dict), axis=1)
+    allsig[["CIS/TRANS","REF_CHR","REF_START","REF_END"]] = pd.DataFrame(cis_tuples.tolist(), index=allsig.index)
+
     try:
         allsig = allsig.where(~pd.isna(allsig), pd.NA)
     except:
@@ -689,6 +691,20 @@ def determine_if_cis(x, group_key,windowsizekb, reference_dict):
     else:
         return "NoReference"
 
+def determine_if_cis2(x, group_key,windowsizekb, reference_dict):
+    if x[group_key] in reference_dict.keys():
+        is_same_chr = str(reference_dict[x[group_key]][0]) == str(x["CHR"])
+        is_large_than_start = int(reference_dict[x[group_key]][1]) - windowsizekb*1000 <= x["POS"]
+        is_smaller_than_end = int(reference_dict[x[group_key]][2]) + windowsizekb*1000 >= x["POS"]               
+        
+        if  is_same_chr and is_large_than_start  and is_smaller_than_end:
+            return "Cis", int(reference_dict[x[group_key]][0]), int(reference_dict[x[group_key]][1]), int(reference_dict[x[group_key]][2])
+        else: 
+            return "Trans", int(reference_dict[x[group_key]][0]), int(reference_dict[x[group_key]][1]), int(reference_dict[x[group_key]][2])
+    else:
+        return "NoReference", pd.NA, pd.NA, pd.NA
+
+
 def determine_distance(allsig, knownsig):
     if len(allsig)==0: 
         return allsig