dag6_seqrun_processing.py 2.5 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788
  1. from datetime import timedelta
  2. from airflow.models import DAG,Variable
  3. from airflow.utils.dates import days_ago
  4. from airflow.operators.bash_operator import BashOperator
  5. from airflow.contrib.operators.ssh_operator import SSHOperator
  6. from airflow.contrib.hooks.ssh_hook import SSHHook
  7. default_args = {
  8. 'owner': 'airflow',
  9. 'depends_on_past': False,
  10. 'start_date': days_ago(2),
  11. 'email_on_failure': False,
  12. 'email_on_retry': False,
  13. 'retries': 1,
  14. 'retry_delay': timedelta(minutes=5),
  15. }
  16. dag = \
  17. DAG(
  18. dag_id='dag6_seqrun_processing',
  19. catchup=False,
  20. schedule_interval="None",
  21. max_active_runs=1,
  22. default_args=default_args)
  23. orwell_ssh_hook = \
  24. SSHHook(
  25. key_file=Variable.get('hpc_ssh_key_file'),
  26. username=Variable.get('hpc_user'),
  27. remote_host='orwell.hh.med.ic.ac.uk')
  28. with dag:
  29. switch_off_project_barcode = \
  30. SSHOperator(
  31. task_id = 'switch_off_project_barcode',
  32. dag = dag,
  33. ssh_hook = orwell_ssh_hook,
  34. queue='hpc_4G',
  35. command = 'bash /home/igf/igf_code/IGF-cron-scripts/orwell/switch_off_project_barcode_check.sh '
  36. )
  37. change_samplesheet_for_run = \
  38. SSHOperator(
  39. task_id = 'change_samplesheet_for_run',
  40. dag = dag,
  41. queue='hpc_4G',
  42. ssh_hook = orwell_ssh_hook,
  43. command = 'bash /home/igf/igf_code/IGF-cron-scripts/orwell/change_samplesheet_for_seqrun.sh '
  44. )
  45. restart_seqrun_processing = \
  46. SSHOperator(
  47. task_id = 'restart_seqrun_processing',
  48. dag = dag,
  49. queue='hpc_4G',
  50. ssh_hook = orwell_ssh_hook,
  51. command = 'bash /home/igf/igf_code/IGF-cron-scripts/orwell/restart_seqrun_processing.sh '
  52. )
  53. register_project_metadata = \
  54. SSHOperator(
  55. task_id = 'register_project_metadata',
  56. dag = dag,
  57. queue='hpc_4G',
  58. ssh_hook = orwell_ssh_hook,
  59. command = 'bash /home/igf/igf_code/IGF-cron-scripts/orwell/register_metadata.sh '
  60. )
  61. find_new_seqrun = \
  62. SSHOperator(
  63. task_id = 'find_new_seqrun',
  64. dag = dag,
  65. queue='hpc_4G',
  66. ssh_hook = orwell_ssh_hook,
  67. command = 'bash /home/igf/igf_code/IGF-cron-scripts/orwell/find_new_seqrun.sh '
  68. )
  69. seed_demultiplexing_pipe = \
  70. BashOperator(
  71. task_id = 'seed_demultiplexing_pipe',
  72. dag = dag,
  73. queue='hpc_4G',
  74. bash_command = 'bash /rds/general/user/igf/home/git_repo/IGF-cron-scripts/hpc/seed_demultiplexing_pipeline.sh '
  75. )
  76. switch_off_project_barcode >> change_samplesheet_for_run >> restart_seqrun_processing
  77. restart_seqrun_processing >> register_project_metadata >> find_new_seqrun