import hail as hl                                                                                            
print("hl.init")                                                                                             
hl.init(default_reference="GRCh38") #log='/home/hail/combiner.log')                                          
print("hl.init done")                                                                                        
                                                                                                             
bucket="gs://interval-wes/gvcf_for_hail_jc/"                                                                 
path_to_input_list = bucket + 'jc/vcf_list.fofn'  # a file with one GVCF path per line                       
print(path_to_input_list)                                                                                    
                                                                                                             
inputs = []                                                                                                  
with hl.hadoop_open(path_to_input_list, 'r') as f:                                                           
    for line in f:                                                                                           
        print("hl.hadoop_open counter:" + str(counter))                                                      
        inputs.append(line.strip())                                                                          
print("hl.hadoop_open done")                                                                                 
                                                                                                             
output_file = bucket + 'jc/interval_wes.combiner.mt'  # output destination                                   
temp_bucket = bucket + 'jc/tmp'  # bucket for storing intermediate files                                     
print(output_file)                                                                                           
print(temp_bucket)          
                        
print("run combiner")                                                                                        
hl.experimental.run_combiner(inputs,                                                                         
                             out_file=output_file,                                                           
                             tmp_path=temp_bucket,                                                           
                             reference_genome='GRCh38')                                                      
print("run combiner done")