dvc.lock

schema: '2.0'
stages:
  run_jackhmmer:
    cmd: python scripts/run_jackhmmer.py
    deps:
    - path: aide_predict/utils/jackhmmer.py
      hash: md5
      md5: 29311d7958e0f3f02f2265184ca0d074
      size: 7534
    - path: data/wt.fasta
      hash: md5
      md5: a1598310bd55723f8c9255b8160c1816
      size: 263
    - path: scripts/run_jackhmmer.py
      hash: md5
      md5: fa97a43d299c54f9bab3bc81c69d2fc2
      size: 1726
    params:
      params.yaml:
        jackhmmer:
          seqdb: /kfs2/projects/bpms/ekomp_tmp/datasets/uniref/uniref100.fasta
          iterations: 5
          evalue: 0.0001
          tvalue: 0.5
          use_bitscores: true
          cpus: 16
          mx: BLOSUM62
          popen: 0.02
          pextend: 0.4
    outs:
    - path: data/jackhmmer/
      hash: md5
      md5: ea7a9ff5cdb4a795b63d838c42e73b50.dir
      size: 42340850998
      nfiles: 2
  run_msa:
    cmd: python scripts/run_msa.py
    deps:
    - path: data/experimental_data.csv
      hash: md5
      md5: d41d8cd98f00b204e9800998ecf8427e
      size: 0
    - path: data/starting_sequences.a2m
      hash: md5
      md5: 624275f98f9f67c5939e3aba6e4a873d
      size: 173269062
    - path: data/starting_sequences.fa
      hash: md5
      md5: d41d8cd98f00b204e9800998ecf8427e
      size: 0
    - path: data/wt.fa
      hash: md5
      md5: 592b18586d4530aaef6557b77c963ee3
      size: 78
    - path: scripts/run_msa.py
      hash: md5
      md5: 0ef6b9083d6ebd257d39cbb9bff52734
      size: 7401
    params:
      params.yaml:
        msa_creation.jackhmmer.cpus: 16
        msa_creation.jackhmmer.domain_threshold: 100
        msa_creation.jackhmmer.iterations: 1
        msa_creation.jackhmmer.mx: BLOSUM62
        msa_creation.jackhmmer.seqdb: uniref100
        msa_creation.jackhmmer.sequence_identity_filter: 0.8
        msa_creation.jackhmmer.sequence_threshold: 100
        msa_creation.jackhmmer.use_bitscores: true
        msa_creation.msa_mode: starting_sequences
        msa_creation.starting_sequences.activity_targets:
        - 0
        msa_creation.starting_sequences.activity_threshold: 0.0
        msa_creation.starting_sequences.add_training_sequences: false
        msa_creation.starting_sequences.prealigned: true
        use_msa: true
    outs:
    - path: data/metrics/run_msa.json
      hash: md5
      md5: 8d205654d9b81dbfb76c9799add5bc43
      size: 61
    - path: data/run_msa/
      hash: md5
      md5: 95ce7fb78568780b4c41bee915e29b69.dir
      size: 173269062
      nfiles: 1
  process_msa:
    cmd: python scripts/process_msa.py
    deps:
    - path: data/run_msa/
      hash: md5
      md5: 95ce7fb78568780b4c41bee915e29b69.dir
      size: 173269062
      nfiles: 1
    - path: data/wt.fa
      hash: md5
      md5: 592b18586d4530aaef6557b77c963ee3
      size: 78
    - path: scripts/process_msa.py
      hash: md5
      md5: 2c829feb65d96f4a77628fe0d42bcfa2
      size: 2433
    params:
      params.yaml:
        msaprocessing.additional_weights: false
        msaprocessing.preprocess: true
        msaprocessing.remove_sequences_with_indeterminate_AA_in_focus_cols: true
        msaprocessing.theta: 0.2
        msaprocessing.threshold_focus_cols_frac_gaps: 0.5
        msaprocessing.threshold_sequence_frac_gaps: 0.5
        msaprocessing.use_weights: true
    outs:
    - path: data/metrics/process_msa.json
      hash: md5
      md5: 3f86e048d9c8f117fdfd3e89c83f1c6b
      size: 83
    - path: data/process_msa/
      hash: md5
      md5: 844813c4a68d4abe187d81facf8c2cbf.dir
      size: 188291636
      nfiles: 2