comparison extract_tables.py @ 14:436da36c6114 draft

Uploaded
author ieguinoa
date Tue, 22 Sep 2020 20:36:41 +0000
parents 91a3a617d94b
children d85e35a36715
comparison
equal deleted inserted replaced
13:91a3a617d94b 14:436da36c6114
32 sample_alias = 'sample_'+str(sample_index) 32 sample_alias = 'sample_'+str(sample_index)
33 samples_table.write('\t'.join([sample_alias,action,'ena_accession',sample['title'],sample['tax_name'], sample['tax_id'],sample['description'],'ENA_submission_date'])+ '\n') 33 samples_table.write('\t'.join([sample_alias,action,'ena_accession',sample['title'],sample['tax_name'], sample['tax_id'],sample['description'],'ENA_submission_date'])+ '\n')
34 for exp_index,exp in enumerate(sample['experiments']): 34 for exp_index,exp in enumerate(sample['experiments']):
35 exp_alias = 'experiment_'+str(exp_index)+'_'+str(sample_index) 35 exp_alias = 'experiment_'+str(exp_index)+'_'+str(sample_index)
36 lib_alias = 'library_'+str(exp_index)+'_'+str(sample_index) 36 lib_alias = 'library_'+str(exp_index)+'_'+str(sample_index)
37 experiments_table.write('\t'.join([exp_alias,action,'accession_ena',exp['title'],study_alias,sample_alias,exp['experiment_design'],lib_alias,exp['library_strategy'],exp['library_source'],exp['library_selection'],exp['library_layout'].lower(),exp['insert_size'],exp['library_construction_protocol'],exp['platform'],exp['instrument_model'],'submission_date_ENA']) + '\n') 37 experiments_table.write('\t'.join([exp_alias,action,'accession_ena',exp['title'],study_alias,sample_alias,exp['experiment_design'],lib_alias,exp['library_strategy'],exp['library_source'],exp['library_selection'],exp['library_layout'],exp['insert_size'],exp['library_construction_protocol'],exp['platform'],exp['instrument_model'],'submission_date_ENA']) + '\n')
38 run_index = 0 38 run_index = 0
39 # exp['runs'] is a list of lists 39 # exp['runs'] is a list of lists
40 for run in exp['runs']: 40 for run in exp['runs']:
41 run_index += 1 41 run_index += 1
42 run_alias = '_'.join(['run',str(exp_index),str(sample_index),str(run_index)]) 42 run_alias = '_'.join(['run',str(exp_index),str(sample_index),str(run_index)])
43 for file_entry in run: 43 for file_entry in run:
44 file_format = 'fastq' 44 file_format = 'fastq.gz' if os.path.splitext(file_entry)[-1] == '.gz' else 'fastq.bz2'
45 runs_table.write('\t'.join([run_alias,action,'ena_run_accession',exp_alias,file_entry,file_format,'file_checksum','submission_date_ENA']) + '\n') 45 runs_table.write('\t'.join([run_alias,action,'ena_run_accession',exp_alias,file_entry,file_format,'file_checksum','submission_date_ENA']) + '\n')
46 46
47 studies_table.close() 47 studies_table.close()
48 samples_table.close() 48 samples_table.close()
49 experiments_table.close() 49 experiments_table.close()