Fix Error in SFS

and change how to write and  load sumstats  files

summary_statistics.py

@@ -369,22 +369,31 @@ def do_sum_stats(scenario_dir, name_id, size_chr=2e6, circular=True, label="", n
             except Exception as e:
                 logging.error("While computing LD for {}\n>>> Error: {}".format(sim_id, e))
 
-    all_sfs2 = all_sfs.groupby("N_indiv").sum() / n_indiv
-    all_sfs2["i_xi_norm"] = all_sfs2.i_xi / all_sfs2.i_xi.sum()
+    all_sfs2 = all_sfs.groupby("N_indiv").mean()
+    all_sfs2["i_xi_norm"] = all_sfs2.i_xi / all_sfs2.i_xi.mean()
     all_sfs2["freq_indiv"] = all_sfs2.index / n_indiv
-    all_sfs2["i_xi_sem_norm"] = all_sfs.groupby("N_indiv").count_SNP.sem() / all_sfs2.i_xi.sum()
+    all_sfs2["i_xi_sem_norm"] = all_sfs.groupby("N_indiv").i_xi.sem() / all_sfs2.i_xi.mean()
     all_sfs2["sim_id"] = sim_id
     all_sfs2["scenario"] = scenario
     all_sfs2["label"] = label
     all_sfs2.reset_index(inplace=True)
-    all_sfs2.to_csv(os.path.join(outdir, name_id + ".sfs"), sep="\t", index=False, mode="a", header=False)
+    writing_mode = "w" if overwrite else "a"
+    with open(os.path.join(outdir, name_id + ".sfs"), writing_mode) as sfsfile:
+        all_sfs2.to_csv(sfsfile,
+                        sep="\t",
+                        index=False,
+                        header=False if (sfsfile.tell()==0 and not overwrite) else True)
 
     all_ld2 = all_ld.groupby("dist_group").mean()
     all_ld2["sim_id"] = sim_id
     all_ld2["scenario"] = scenario
     all_ld2["label"] = label
     all_ld2.reset_index(inplace=True)
-    all_ld2.to_csv(os.path.join(outdir, name_id + ".ld"), sep="\t", index=False, mode="a", header=False)
+    with open(os.path.join(outdir, name_id + ".ld"), writing_mode) as ldfile:
+        all_ld2.to_csv(ldfile,
+                       sep="\t",
+                       index=False,
+                       header=False if (ldfile.tell()==0 and not overwrite) else True)
 
 
     #
@@ -417,12 +426,9 @@ def do_sum_stats(scenario_dir, name_id, size_chr=2e6, circular=True, label="", n
 def load_sum_stats(name_id, label=""):
     """Load data from name_id/name_id.{sfs|ld|sel} and return the 3 df"""
 
-    df_sfs = pd.read_table(os.path.join(name_id+ ".sfs"),
-                           header=None,
-                           names=["N_indiv", "count_SNP", "i_xi", "freq_indiv", "i_xi_norm", "i_xi_sem_norm", "sim_id", "scenario", "label"])
-    df_ld = pd.read_table(os.path.join(name_id+ ".ld"),
-                          header=None,
-                          names=["dist_group", "mean_dist", "mean_r2", "Count", "sem_r2", "sim_id", "scenario", "run_id", "label"])
+    df_sfs = pd.read_table(os.path.join(name_id+ ".sfs"))
+    df_ld = pd.read_table(os.path.join(name_id+ ".ld"))
+
     # df_sel = pd.read_table(os.path.join(name_id+ ".sel"),
     #                        header=None,
     #                        names=["position_percent", "TajD", "IHS", "NSL", "sim_id", "scenario", "run_id", "label"])