config.json

{

  "regions":{
      "start":[1000000],
      "end":[10000000]
  },
  "regions_files":["/home/a/ag568/region_0.txt"],
  "chromosome": "chr1",
  "reference_file":{
    "filename": "/scratch/spectre/a/ag568/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna"
  },
  "no_wga_file": {
     "filename": "/scratch/spectre/a/ag568/m585_verysensitive_trim_sorted.bam"
  },
  "test_file":{
    "filename": "/scratch/spectre/a/ag568/m605_verysensitive_trim_sorted.bam"
  },
  "debug":{
    "log_bam_for_debug":false
  },
  "window_size":100,
  "fill_missing_window_data":false,
  "fill_missing_window_data_factor":0,
  "quality_threshold":null,
  "save_windows":false,
  "remove_windows_with_N":false,
  "mark_N_windows":true,
  "mark_for_N_windows":-999.0,
  "save_cluster_dbi":true,
  "n_windows_dist":{
    "name":"uniform",
    "config":{
      "parameters":[-999.5, -998.5]
    }
  },
  "windows_filename":"hmm_windows.json",
  "check_windowing_sanity":true,
  "outlier_remove":{

    "name":"zscore",
    "config":{
      "sigma_factor":2
    }
  },
  "label_clusters": true,
  "labeler":{
    "name":"mean_diff",
    "tuf_mean_min": 1.5,
    "tuf_mean_max": 8.5,
    "states":["DELETE", "OTHER", "OTHER", "TUF"]
  },
  "clusterer":{
    "name":"kmedoids",
    "config":{
    "init_cluster_idx":"random_from_data",
    "metric":"MANHATAN",
    "features":["mean"],
    "n_clusters":4
    }
  },
  "clusters":{
      "cluster_0":{
        "filename":"/home/a/ag568/cluster_0.txt",
        "state": "normal",
         "distributions":{
          "wga":{
            "type":"distribution",
            "name":"normal"
            },
            "no_wga":{
              "type":"distribution",
              "name":"normal"
            }
        }
      },
      "cluster_1":{
        "filename":"/home/a/ag568/cluster_1.txt",
        "state": "tuf",
        "distributions":{
          "wga":{
            "type":"gmm",
            "uniform":{
              "params":[2.0, 8.0]
            },
            "names":["normal", "uniform"],
            "weights":null
          },
          "no_wga":{
            "type":"gmm",
            "uniform":{
              "params":[2.0, 8.0]
            },
            "names":["normal", "uniform"],
            "weights":null
          }
        }
      }
  },

  "HMM": {
    "name":"HMM_Model",
    "train":true,
    "train_solver": "baum-welch",
    "lr_decay":0.7,
    "inertia":0.6,
    "verbose":true,
    "save_model":true,
    "save_hmm_filename":"/home/a/ag568/HMM_Model",
    "states":{
      "tuf":{
         "start_prob":0.48
      },
      "normal":{
         "start_prob":0.48
      },
      "gap_state":{
          "start_prob":0.04
      }
    },
    "transitions":{
        "tuf-tuf":0.95,
        "normal-normal":0.95,
        "tuf-normal":0.05,
        "normal-tuf":0.05,
        "gap_state-gap_state":0.95,
        "gap_state-tuf":0.05,
        "tuf-gap_state":0.05,
        "normal-gap_state":0.05,
        "gap_state-normal":0.05
     },
    "train_sequence_size":100,
    "train_sequence_source":"region",
    "train_n_sequences_per_source":10000,
    "train_windowtype":"both"
  },
  "logger_file":"tuf.log",
  "logger_level":"INFO"
}