{
  "study": "mfass-specialist-annotation-matched-sensitivity",
  "protocol": "docs/mfass-specialist-comparison.md",
  "registration": "docs/mfass-matched-study-registration.md",
  "registration_sha256": "7f03c13d3c7fa73ffe91d3170fe51f06a277d34b5745476ce45ca8a7cece9162",
  "frozen_at_utc": "2026-09-24T19:00:42+00:00",
  "draft": false,
  "status": "frozen; the output directory was empty when this manifest was written",
  "git": {
    "head": "b52a3daef6e85ee869e70732a586355f83dc30fb",
    "dirty_paths": []
  },
  "runner_files_sha256": {
    "benchmarks/mfass/src/mfass/run_spliceai.py": "3d238f11069049cbae08aa779f2e0857b1f8d2929f4e03a233848b6d5f06c7ba",
    "benchmarks/mfass/src/mfass/run_pangolin.py": "13ba2ca1d258ca7482f9d627155402284e4ec2324ac04e98876be8df5976dc63",
    "benchmarks/mfass/src/mfass/checkpoint.py": "dcd271576ec49dba23b9fc30620267de88f7620a647b0a6c53b1a295dc21ecc0",
    "benchmarks/mfass/src/mfass/compare.py": "53e88e6bb4fff70f202aaaaa0a328535dac33905a418270054475abc3b26ec17",
    "benchmarks/mfass/src/mfass/specialist_run.py": "242cfaa770f0ae5e27696fe5e2e8d9e8068b5de457a4f0bd0a89dda31075b4de",
    "benchmarks/mfass/src/mfass/specialist_provenance.py": "0a8bb78598a583b894d3c48b7aa442cd287d5d4de9b987ce60ad9622ca5f14db",
    "benchmarks/mfass/src/mfass/pangolin_patch.py": "4bb77dde473b8d535a323d92b3d71f8d96c681af766e8419f0d9c7f8d3bee8c6",
    "benchmarks/mfass/src/mfass/matched_annotation.py": "62d76225d324d249a3d17e624628a83757b31b209e4fe2999b87a874f034df08",
    "benchmarks/mfass/src/mfass/matched_study.py": "f8782703da70962b5623d467a0cc1505412d409593c68ac765f1a870f50db1b2",
    "packages/rewirebench/src/rewirebench/metrics.py": "4030374ed11d38e523fc9991ddd792f8800e7739ac71588085bfff91f912d7f7",
    "packages/rewirebench/src/rewirebench/results.py": "e3b95804fb9cd898a125e78d3abad8022df186d094b89ac0c8588070af4a516d",
    "packages/rewirebench/src/rewirebench/protocols/mfass.py": "6d090a6b1622fe571b8c48ecc213b9b6adf3cbf06cc2abc2d3bfc891a3f8ed93",
    "benchmarks/mfass/patches/pangolin-5cf94b8-mask-per-gene-1.patch": "d1c9fb69245f9d256e2547fd8bf61547686cfc3caf6c40c1351ab2d6997cb454",
    "benchmarks/mfass/scripts/run_matched_study.sh": "697c77056ba16bc0c36235f6aac68986f1fe0a3abaa1ce15801e6e3aa788b481",
    "uv.lock": "3f2ec7f464119ec2478ad15aac302f1ba1e16f77725e7d30ffb2af68bf5f4296"
  },
  "paths": {
    "cohort": "$MFASS_DATA/cohort.tsv",
    "split": "$REPO/benchmarks/mfass/splits/split-v2.tsv",
    "fasta": "$STUDY_DATA/ref/GRCh38.primary_assembly.genome.fa",
    "fai": "$STUDY_DATA/ref/GRCh38.primary_assembly.genome.fa.fai",
    "spliceai_annotation": "$STUDY_DATA/annotation/gencode44-canonical-v1/spliceai.tsv",
    "pangolin_db": "$STUDY_DATA/annotation/gencode44-canonical-v1/pangolin.db",
    "annotation_release": "GENCODE v44 primary assembly (MD5 checked against official MD5SUMS)"
  },
  "resources_sha256": {
    "cohort": "389702ff4c647d7ce10a90092a6fa811ae777d15997baf39ce9aae0346247bd0",
    "split": "999ebcb7e63a5c5eaa8780fa468e59ac1f934260ad50102814174c396317f052",
    "fasta": "e49b92b3e4f321bf254c042f25b726d9931c4d74c7523e8b6bb530e63b0cfd4b",
    "fai": "a2c323ea4cff34d7123ace4578f7e122b2d2f5a22f40dc23eb8b97d17723d169",
    "spliceai_annotation": "2383cd148b292b9a6af56da160c9c142ddeb5720b1e9ef7c5966a1744f684c4d",
    "pangolin_db": "84ae1a18d9d7a32e14e3fe9c8705256a28cd9bde0a55bc46200cbe77178deb27"
  },
  "annotation": {
    "summary_sha256": "0921c19d1f6a71c4afaf94f366dfbdf2e87fb3435e8a7389493c0d99c04de7f8",
    "cross_check_sha256": "71caa6d5d2e2edfc88bb5bce2dc35a2abd0d0c926c56cf85ffe6eafbee2ee086",
    "outputs_sha256": {
      "selection.tsv": "805726643430ea4625b5bb86b8081ce65f3bddbca57868ba44bbadff08dd0c72",
      "exclusions.tsv": "110c875c337a8a9ff8904997e93eb768e3d85a44a58ff6c09f7d23d93afd0478",
      "spliceai.tsv": "2383cd148b292b9a6af56da160c9c142ddeb5720b1e9ef7c5966a1744f684c4d",
      "pangolin.gtf": "c999d9bc14cf3d05a5994915e32a7e4e0ae2d020cee8b5578f89b2defffb1edc",
      "cross-check.json": "71caa6d5d2e2edfc88bb5bce2dc35a2abd0d0c926c56cf85ffe6eafbee2ee086",
      "pangolin.db": "84ae1a18d9d7a32e14e3fe9c8705256a28cd9bde0a55bc46200cbe77178deb27"
    }
  },
  "eligibility": {
    "summary_sha256": "ec2c57b5ac88ca4011ff02351e4a20092e80d7de75f81acbe8ffb00851ddff93",
    "test_variants": 8324,
    "checked": 8324,
    "gene_sets_differ": 0,
    "multiple_genes_same_strand_pangolin": 344,
    "pangolin_has_gene": 8320,
    "reference_mismatch_pangolin_rule": 23,
    "reference_mismatch_spliceai_rule": 23,
    "scoreable_pangolin": 8297,
    "scoreable_spliceai": 8297,
    "spliceai_has_gene": 8320,
    "labels_read": false,
    "annotation_dir_summary_sha256": "0921c19d1f6a71c4afaf94f366dfbdf2e87fb3435e8a7389493c0d99c04de7f8",
    "pangolin_source_sha256": "239c6ba9b1883d712d4856985c6eaa07e5b67f99790a718f93d6316ad076c083",
    "pangolin_source_identity": "pangolin-5cf94b8-mask-per-gene-1",
    "cohort_sha256": "389702ff4c647d7ce10a90092a6fa811ae777d15997baf39ce9aae0346247bd0",
    "split_sha256": "999ebcb7e63a5c5eaa8780fa468e59ac1f934260ad50102814174c396317f052",
    "fasta_sha256": "e49b92b3e4f321bf254c042f25b726d9931c4d74c7523e8b6bb530e63b0cfd4b",
    "tool_versions": {
      "python": "3.11.13",
      "sqlite": "3.49.1",
      "gffutils": "0.14",
      "pandas": "2.3.3",
      "numpy": "1.26.4"
    },
    "output_sha256": "a0b1a01c8be33e95bc1e0635a30c0f439f8eb732527982b3db13a194989a66a5"
  },
  "code": {
    "spliceai": {
      "spliceai_version": "1.3.1",
      "spliceai_files": {
        "__init__.py": {
          "sha256": "5e0ade84087ee01040dbc3344359cbfdb26e2fcf98c214e15110ac1d05562037",
          "git_blob": "b5d843e672cfff52c6d050205dead13d7c31533e"
        },
        "utils.py": {
          "sha256": "207e147c73bdc1970fdb176dcbf088883f9a58756bfeb0d93494e75018ae477c",
          "git_blob": "66f2c3c14198d9020a7913f0412fc91ade008b91"
        },
        "__main__.py": {
          "sha256": "1090e97a5f7d8f937048b7d1f08d98fb593bd86953e9d64178aa3ca096961a77",
          "git_blob": "2c59acd61ce528dd0ed6d6c14244579354fed146"
        },
        "models/spliceai1.h5": {
          "sha256": "e1fd5adcef7489d604b10e79c40078ef790d51ef048c4ce3869c9119ac5de42b",
          "git_blob": "e5e6bf5d2f61f9f9f8f1c93386881992a96a82c0"
        },
        "models/spliceai2.h5": {
          "sha256": "6ab042b82ab966b6d3582cb31b96f0859ea08a864f168d69e83aa14450a3b66e",
          "git_blob": "b6c197b52fc39e81f42a2dfb25eeef11db3c94af"
        },
        "models/spliceai3.h5": {
          "sha256": "e2e790bde53dfdf410c6dc434a86122a7d12f3f38dc2ef45d85986e9ecf22fad",
          "git_blob": "5009694c892a782faa70f66e23f0f74cf69c6809"
        },
        "models/spliceai4.h5": {
          "sha256": "ca88ac9e58e69ba6fdeed319b72f063f164c9abf7392eaccef903e94c1d99dd6",
          "git_blob": "81ce1c887855f34abf9af5f5c65d85fd04d3132d"
        },
        "models/spliceai5.h5": {
          "sha256": "791cd22c62a80a08d2ca674615a93ce8159d7b55bd157cfef2983b1bd6b41391",
          "git_blob": "752679198e4f52f69b93945b2c49e4efc8534860"
        }
      },
      "spliceai_wheel_only_files_sha256": {
        "__clean_main__.py": "d91e273a60b1b1c079cd29974f1c9901d997bc6f57226ff094ad15d0c1e49b03",
        "normalise_chrom.py": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b"
      },
      "spliceai_source_identity": "upstream-b3c7f17",
      "spliceai_upstream_files_verified": true,
      "spliceai_upstream_file_mismatches": []
    },
    "pangolin": {
      "pangolin_source_sha256": "239c6ba9b1883d712d4856985c6eaa07e5b67f99790a718f93d6316ad076c083",
      "pangolin_source_identity": "pangolin-5cf94b8-mask-per-gene-1",
      "pangolin_version": "1.0.2+rewire.maskpergene1",
      "pangolin_patch": {
        "id": "pangolin-5cf94b8-mask-per-gene-1",
        "patch_sha256": "d1c9fb69245f9d256e2547fd8bf61547686cfc3caf6c40c1351ab2d6997cb454",
        "upstream_revision": "5cf94b8db938c658391b4305cd7ce33297d44ff7",
        "upstream_source_sha256": "f3f25c4febf64d01ef42f967dc5cf10f6856bf3aa92d26f09803ad408169da9b"
      },
      "pangolin_files": {
        "__init__.py": {
          "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
          "git_blob": "e69de29bb2d1d6434b8b29ae775ad8c2e48c5391"
        },
        "model.py": {
          "sha256": "4a1c5c2570aafe1452bb43332255321677e6c6c817adf84b9dd438e3ca4be6f8",
          "git_blob": "11dfb436b36f9b5783874b2db636a1788bfab619"
        },
        "models/final.1.0.3.v2": {
          "sha256": "f0478fab173b75f7f7e9fe96688bad6c50fa4a46d70557f423b110caaf565501",
          "git_blob": "fbe03218df2f781a9713b05bfa42001ae548782a"
        },
        "models/final.2.0.3.v2": {
          "sha256": "c4c6bb4880fa6fb28b14182ae3ea0600edb07056158f55325b5e6e6e48fc9f26",
          "git_blob": "ec4ee767bc83fa0397222f42c0da210a47f13fbb"
        },
        "models/final.3.0.3.v2": {
          "sha256": "ec685a6e7105a4486c1f89a005458a13deb3fe7171f13d434f4877e386d10676",
          "git_blob": "f31cbae821e38bc4361216c8e87c8db13fd5a9c0"
        },
        "models/final.1.2.3.v2": {
          "sha256": "559c05de3e1ce65c2515ca3e92ef85edb0ec2e47686ca58060e25891ce06eb3a",
          "git_blob": "39e8756faf1e427cd9a8c6c691532b3683d8916a"
        },
        "models/final.2.2.3.v2": {
          "sha256": "48758ba8b95eee9aa9feea52672ef06ca1b34111299c27f8a710f734d8b9aae5",
          "git_blob": "6fd0a31789c7de788227e103f33d697ad538def6"
        },
        "models/final.3.2.3.v2": {
          "sha256": "7cb576c2b24db4fdd6970c4ca4fb7c20ae1b1d8ae80645ebbe689848b5743129",
          "git_blob": "c7f2ae10caab8f5bae167309b2b7d3e6905871e5"
        },
        "models/final.1.4.3.v2": {
          "sha256": "c50b12e0c0af776d5674ca5e346493f8265783494d4df383364de9c1136657f6",
          "git_blob": "0af4fadb3f3dd54c49d41042befefa19fd4bd32a"
        },
        "models/final.2.4.3.v2": {
          "sha256": "e03303bed4fd6f135ec0f6c1b192cce954ea42d0646f44d17b4a6fbb2b1f610e",
          "git_blob": "8c003df629c9f1e8072000f8bbc3eeab94850b6c"
        },
        "models/final.3.4.3.v2": {
          "sha256": "9476d2e25520d7ff15bece0cd5d3b657e3b1dd3cc5fcab1d9c3b62bea7a0c5b6",
          "git_blob": "9ed4fad50ead41761073cdff3d9acafc6f78bdd4"
        },
        "models/final.1.6.3.v2": {
          "sha256": "2aae563fa18a8a9b6699c6c96e0d32b8ec7543f8f805fb3bc9de77302cc9f66e",
          "git_blob": "98e5539fe339385e9c8b19d2786c4f8c3f186a3b"
        },
        "models/final.2.6.3.v2": {
          "sha256": "7d3c0b1b2a60067b940dec315567874fbc8bcd322f1b7c76bf969f51f0f53f7f",
          "git_blob": "6a650d629fae6aee8171708241a0f580bb368cfd"
        },
        "models/final.3.6.3.v2": {
          "sha256": "756e7721a382cace24e9bfea5b543af5623f2487d9a3efe7385e9c76367005fd",
          "git_blob": "4a8e9e33d5832770e6563d0229bdb4bcae5b8398"
        }
      },
      "pangolin_upstream_files_verified": true,
      "pangolin_upstream_file_mismatches": []
    }
  },
  "environments": {
    "spliceai": {
      "python": "$STUDY_DATA/envs/spliceai/bin/python",
      "distributions": [
        "Pygments==2.21.0",
        "absl-py==2.5.0",
        "astunparse==1.6.3",
        "certifi==2026.7.22",
        "charset-normalizer==3.5.1",
        "cloudpickle==3.1.2",
        "flatbuffers==25.12.19",
        "gast==0.7.0",
        "google-pasta==0.2.0",
        "grpcio==1.84.0",
        "h5py==3.14.0",
        "idna==3.19",
        "iniconfig==2.3.0",
        "joblib==1.6.0",
        "keras==3.15.1",
        "libclang==18.1.1",
        "markdown-it-py==4.2.0",
        "mdurl==0.1.2",
        "mfass==0.1.0",
        "ml_dtypes==0.5.4",
        "namex==0.1.0",
        "narwhals==2.26.0",
        "numpy==1.26.4",
        "opt_einsum==3.4.0",
        "optree==0.20.0",
        "packaging==26.3",
        "pandas==2.3.3",
        "pluggy==1.6.0",
        "protobuf==7.36.1",
        "pyfaidx==0.9.0.4",
        "pysam==0.24.1",
        "pytest==9.1.1",
        "python-dateutil==2.9.0.post0",
        "pytz==2026.3.post1",
        "requests==2.34.2",
        "rewirebench==0.5.0",
        "rich==15.0.0",
        "ruff==0.16.7",
        "scikit-learn==1.9.1",
        "scipy==1.17.1",
        "setuptools==81.0.0",
        "six==1.17.0",
        "spliceai==1.3.1",
        "tensorflow==2.21.0",
        "termcolor==3.3.0",
        "threadpoolctl==3.6.0",
        "typing_extensions==4.16.0",
        "tzdata==2026.4",
        "urllib3==2.7.0",
        "wheel==0.48.0",
        "wrapt==2.4.1"
      ],
      "distributions_sha256": "8f0eb1412c265d12f505332f2900c44c03fcc7ba64bcdbe0afb94833cd66302f"
    },
    "pangolin": {
      "python": "$STUDY_DATA/envs/pangolin/bin/python",
      "distributions": [
        "Jinja2==3.1.6",
        "MarkupSafe==3.0.3",
        "PyVCF3==1.0.4",
        "Pygments==2.21.0",
        "argcomplete==3.7.2",
        "argh==0.31.3",
        "cloudpickle==3.1.2",
        "filelock==3.32.6",
        "fsspec==2026.7.0",
        "gffutils==0.14",
        "iniconfig==2.3.0",
        "joblib==1.6.0",
        "mfass==0.1.0",
        "mpmath==1.3.0",
        "narwhals==2.26.0",
        "networkx==3.6.1",
        "numpy==1.26.4",
        "packaging==26.3",
        "pandas==2.3.3",
        "pangolin==1.0.2+rewire.maskpergene1",
        "pluggy==1.6.0",
        "pyfaidx==0.9.0.4",
        "pyfastx==2.3.1",
        "pytest==9.1.1",
        "python-dateutil==2.9.0.post0",
        "pytz==2026.3.post1",
        "rewirebench==0.5.0",
        "ruff==0.16.7",
        "scikit-learn==1.9.1",
        "scipy==1.17.1",
        "setuptools==81.0.0",
        "simplejson==4.1.2",
        "six==1.17.0",
        "sympy==1.14.0",
        "threadpoolctl==3.6.0",
        "torch==2.2.2",
        "typing_extensions==4.16.0",
        "tzdata==2026.4"
      ],
      "distributions_sha256": "3631d391fa7c2f78cf0469fcb59a068f893e50e1bc521e36e57749fb19be9529"
    }
  },
  "threads": {
    "spliceai": {
      "intra_op": 5,
      "inter_op": 1,
      "env": {
        "OMP_NUM_THREADS": "5",
        "MKL_NUM_THREADS": "5",
        "OPENBLAS_NUM_THREADS": "5",
        "VECLIB_MAXIMUM_THREADS": "5",
        "NUMEXPR_NUM_THREADS": "5"
      }
    },
    "pangolin": {
      "intra_op": 6,
      "inter_op": 1,
      "env": {
        "OMP_NUM_THREADS": "6",
        "MKL_NUM_THREADS": "6",
        "OPENBLAS_NUM_THREADS": "6",
        "VECLIB_MAXIMUM_THREADS": "6",
        "NUMEXPR_NUM_THREADS": "6"
      }
    },
    "other_steps": {
      "env": {
        "OMP_NUM_THREADS": "6",
        "MKL_NUM_THREADS": "6",
        "OPENBLAS_NUM_THREADS": "6",
        "VECLIB_MAXIMUM_THREADS": "6",
        "NUMEXPR_NUM_THREADS": "6"
      }
    },
    "cpu_threads": 6,
    "scope": "One step runs at a time. TensorFlow intra-op plus inter-op pools are within the CPU budget. Torch's intra-op pool is within the budget; its inter-op pool is 1 and idle during sequential forward passes. Runtime I/O and Python threads are not counted. Manifest hashing in the parent process is outside all recorded timings."
  },
  "distance": 50,
  "conditions": {
    "S0": {
      "tool": "spliceai",
      "mask": "0",
      "stem": "spliceai-1.3.1-gencode44-canonical-mask0"
    },
    "S1": {
      "tool": "spliceai",
      "mask": "1",
      "stem": "spliceai-1.3.1-gencode44-canonical-mask1"
    },
    "P0": {
      "tool": "pangolin",
      "mask": "False",
      "stem": "pangolin-gencode44-canonical-maskFalse"
    },
    "P1": {
      "tool": "pangolin",
      "mask": "True",
      "stem": "pangolin-gencode44-canonical-maskTrue"
    }
  },
  "contrasts": [
    {
      "name": "S1-S0",
      "candidate": "S1",
      "baseline": "S0"
    },
    {
      "name": "P1-P0",
      "candidate": "P1",
      "baseline": "P0"
    },
    {
      "name": "P0-S0",
      "candidate": "P0",
      "baseline": "S0"
    }
  ],
  "analysis": {
    "capacity": 100,
    "draws": 2000,
    "seed": 20260914,
    "min_common": 100,
    "interval": "95% percentile, whole-group",
    "single_class_refusal": "more than 5% of draws"
  },
  "sequence": [
    "S0",
    "S1",
    "P0",
    "P1",
    "verify",
    "S1-S0",
    "P1-P0",
    "P0-S0",
    "report"
  ],
  "out_dir": "$STUDY_DATA/runs/v1",
  "steps": {
    "S0": [
      "$STUDY_DATA/envs/spliceai/bin/python",
      "-m",
      "mfass.run_spliceai",
      "--annotation",
      "$STUDY_DATA/annotation/gencode44-canonical-v1/spliceai.tsv",
      "--inter-threads",
      "1",
      "--cohort",
      "$MFASS_DATA/cohort.tsv",
      "--split",
      "$REPO/benchmarks/mfass/splits/split-v2.tsv",
      "--ref",
      "$STUDY_DATA/ref/GRCh38.primary_assembly.genome.fa",
      "--annotation-release",
      "GENCODE v44 primary assembly (MD5 checked against official MD5SUMS)",
      "--distance",
      "50",
      "--mask",
      "0",
      "--threads",
      "5",
      "--out",
      "$STUDY_DATA/runs/v1/S0/spliceai-1.3.1-gencode44-canonical-mask0.json",
      "--checkpoint",
      "$STUDY_DATA/runs/v1/S0/checkpoint.tsv",
      "--require-verified-code"
    ],
    "S1": [
      "$STUDY_DATA/envs/spliceai/bin/python",
      "-m",
      "mfass.run_spliceai",
      "--annotation",
      "$STUDY_DATA/annotation/gencode44-canonical-v1/spliceai.tsv",
      "--inter-threads",
      "1",
      "--cohort",
      "$MFASS_DATA/cohort.tsv",
      "--split",
      "$REPO/benchmarks/mfass/splits/split-v2.tsv",
      "--ref",
      "$STUDY_DATA/ref/GRCh38.primary_assembly.genome.fa",
      "--annotation-release",
      "GENCODE v44 primary assembly (MD5 checked against official MD5SUMS)",
      "--distance",
      "50",
      "--mask",
      "1",
      "--threads",
      "5",
      "--out",
      "$STUDY_DATA/runs/v1/S1/spliceai-1.3.1-gencode44-canonical-mask1.json",
      "--checkpoint",
      "$STUDY_DATA/runs/v1/S1/checkpoint.tsv",
      "--require-verified-code"
    ],
    "P0": [
      "$STUDY_DATA/envs/pangolin/bin/python",
      "-m",
      "mfass.run_pangolin",
      "--db",
      "$STUDY_DATA/annotation/gencode44-canonical-v1/pangolin.db",
      "--interop-threads",
      "1",
      "--cohort",
      "$MFASS_DATA/cohort.tsv",
      "--split",
      "$REPO/benchmarks/mfass/splits/split-v2.tsv",
      "--ref",
      "$STUDY_DATA/ref/GRCh38.primary_assembly.genome.fa",
      "--annotation-release",
      "GENCODE v44 primary assembly (MD5 checked against official MD5SUMS)",
      "--distance",
      "50",
      "--mask",
      "False",
      "--threads",
      "6",
      "--out",
      "$STUDY_DATA/runs/v1/P0/pangolin-gencode44-canonical-maskFalse.json",
      "--checkpoint",
      "$STUDY_DATA/runs/v1/P0/checkpoint.tsv",
      "--require-verified-code"
    ],
    "P1": [
      "$STUDY_DATA/envs/pangolin/bin/python",
      "-m",
      "mfass.run_pangolin",
      "--db",
      "$STUDY_DATA/annotation/gencode44-canonical-v1/pangolin.db",
      "--interop-threads",
      "1",
      "--cohort",
      "$MFASS_DATA/cohort.tsv",
      "--split",
      "$REPO/benchmarks/mfass/splits/split-v2.tsv",
      "--ref",
      "$STUDY_DATA/ref/GRCh38.primary_assembly.genome.fa",
      "--annotation-release",
      "GENCODE v44 primary assembly (MD5 checked against official MD5SUMS)",
      "--distance",
      "50",
      "--mask",
      "True",
      "--threads",
      "6",
      "--out",
      "$STUDY_DATA/runs/v1/P1/pangolin-gencode44-canonical-maskTrue.json",
      "--checkpoint",
      "$STUDY_DATA/runs/v1/P1/checkpoint.tsv",
      "--require-verified-code"
    ],
    "S1-S0": [
      "$STUDY_DATA/envs/pangolin/bin/python",
      "-m",
      "mfass.compare",
      "--baseline",
      "$STUDY_DATA/runs/v1/S0/spliceai-1.3.1-gencode44-canonical-mask0.predictions.tsv",
      "--candidate",
      "$STUDY_DATA/runs/v1/S1/spliceai-1.3.1-gencode44-canonical-mask1.predictions.tsv",
      "--capacity",
      "100",
      "--draws",
      "2000",
      "--seed",
      "20260914",
      "--min-common",
      "100",
      "--out",
      "$STUDY_DATA/runs/v1/contrasts/S1-S0.json"
    ],
    "P1-P0": [
      "$STUDY_DATA/envs/pangolin/bin/python",
      "-m",
      "mfass.compare",
      "--baseline",
      "$STUDY_DATA/runs/v1/P0/pangolin-gencode44-canonical-maskFalse.predictions.tsv",
      "--candidate",
      "$STUDY_DATA/runs/v1/P1/pangolin-gencode44-canonical-maskTrue.predictions.tsv",
      "--capacity",
      "100",
      "--draws",
      "2000",
      "--seed",
      "20260914",
      "--min-common",
      "100",
      "--out",
      "$STUDY_DATA/runs/v1/contrasts/P1-P0.json"
    ],
    "P0-S0": [
      "$STUDY_DATA/envs/pangolin/bin/python",
      "-m",
      "mfass.compare",
      "--baseline",
      "$STUDY_DATA/runs/v1/S0/spliceai-1.3.1-gencode44-canonical-mask0.predictions.tsv",
      "--candidate",
      "$STUDY_DATA/runs/v1/P0/pangolin-gencode44-canonical-maskFalse.predictions.tsv",
      "--capacity",
      "100",
      "--draws",
      "2000",
      "--seed",
      "20260914",
      "--min-common",
      "100",
      "--out",
      "$STUDY_DATA/runs/v1/contrasts/P0-S0.json"
    ]
  }
}