Add a ld_score to ensure that imputed SNPs have been imputed from enough typed SNPs

doc/source/index.rst

@@ -88,7 +88,6 @@ Output
 ======
 
 The raiss package outputs imputed GWAS files in the tabular format:
-..TODO suppress complementary columns
 
 +-------------+----------+------------+------------+---------+-------+----------+
 |   snp_ids   | position | Ref_allele | Alt_allele | z_score |  Var  | ld_score |

raiss/filter_format_output.py

@@ -4,7 +4,7 @@
 """
 
 
-def filter_output(zscores, fout, R2_threshold = 0.6):
+def filter_output(zscores, fout, R2_threshold = 0.6, minimum_ld = 20):
     """
     procedure that format output for JASS
 
@@ -18,6 +18,7 @@ def filter_output(zscores, fout, R2_threshold = 0.6):
 
     zscores.reset_index(inplace = True)
     chr_fo = zscores[['index', 'pos', 'A0', 'A1', 'Z', 'Var', "ld_score"]]
-    chr_fo.columns = ['rsID','pos','A0','A1','Z', 'Var', "ld_score"]
+    chr_fo["imputation_R2"] = 1-chr_fo["Var"]
+    chr_fo.columns = ['rsID','pos','A0','A1','Z', 'Var', "ld_score", "imputation_R2"]
 
-    chr_fo.loc[chr_fo.Var < (1-R2_threshold)].to_csv(fout, sep="\t", index=False)
+    chr_fo.loc[(chr_fo.Var < (1-R2_threshold)) & (chr_fo.ld_score > minimum_ld)].to_csv(fout, sep="\t", index=False)

raiss/pipes.py

@@ -41,6 +41,7 @@ def save_chromosome_imputation(gwas, chrom, window_size, buffer_size,
 
     # Formatting and filtering
     # and Saving results
+    minimum_ld = 20
     z_fo = "{0}/z_{1}_{2}{3}.txt".format(output_folder, gwas, chrom, tag)
     filter_output(imputed_zscore, z_fo, float(R2_threshold))
     print("Save imputation done at {0}".format(z_fo))

raiss/stat_models.py

@@ -53,8 +53,7 @@ def compute_var(sig_i_t, sig_t_inv, lamb, batch=True):
     return var, ld_score
 
 def check_inversion(sig_t, sig_t_inv):
-    print(sig_t)
-    print(sig_t_inv)
+
     return np.allclose(sig_t, np.dot(sig_t, np.dot(sig_t_inv, sig_t)))
 
 def var_in_boundaries(var,lamb):