parameters_1kb.yaml

env: "cluster"

min_dist: 1000
max_dist: 100000
window_size: 1000

seed: 413666

training_data_size: 50000 
tuning_data_size: 5000 
num_random_search: 10
ensemble_size: 5

num_dnasechipseq_experiments: 3337
num_chromhmm_experiments: 127
num_chromhmm_states: 25

paths:
  dnasechipseq_experiment_paths_file: "/u/project/ernst/skwon94/pairwise_scoring/files_hg38_DNaseChIPseq.txt"
  chromhmm_experiment_paths_file: "/u/project/ernst/skwon94/pairwise_scoring/files_hg38_ChromHMM.txt"
  chrom_size_file: "/u/project/ernst/skwon94/pairwise_scoring/hg38.chrom.sizes"
  output_dir: "/u/project/ernst/skwon94/pairwise_scoring/model/gw_1kb"

# Peak call data from DNase and ChIP-seq: 3,337 DNase-seq and TF and histone mark ChIP-seq experiments from ENCODE4 and Roadmap with hg38 as genome assembly
# ChromHMM chromatin state annotation: 25-state model learned from imputed data for 127 epigenomes from Roadmap (lifted over from hg19 to hg38)