dvc.yaml

stages:
  download:
    cmd: python gtex.py download
    outs:
      - data/gtex/input/GTEx_8_tissues_snRNAseq_atlas_071421.public_obs.h5ad
      - data/gtex/input/GTEx_8_tissues_snRNAseq_immune_atlas_071421.public_obs.h5ad
      - data/gtex/interim/GTEx_Analysis_2017-06-05_v8_RNASeQCv1.1.9_gene_tpm_stable_expressions.tsv
      - data/gtex/input/GTEx_Analysis_2017-06-05_v8_RNASeQCv1.1.9_gene_tpm_stable.gct
      - data/gtex/input/GTEx_Analysis_v8_Annotations_SampleAttributesDS.txt
      - data/gtex/input/GTEx_Analysis_v8_Annotations_SubjectPhenotypesDS.txt
  prepare_cattle:
    cmd: python gtex.py  prepare-cattle-bulk
    outs:
      - data/cattle/input/Gene_read_counts_FarmGTEx_cattle_V0.tsv
  prepare_bulk:
    deps:
      - data/gtex/input/GTEx_Analysis_v8_Annotations_SampleAttributesDS.txt
      - data/gtex/input/GTEx_Analysis_v8_Annotations_SubjectPhenotypesDS.txt
    cmd: python gtex.py prepare-bulk
    outs:
      - data/gtex/interim/samples_with_subjects.parquet
      - data/gtex/interim/expressions_extended.parquet
  train_gtex:
    cmd: python gtex.py train
    deps:
      - data/gtex/interim/expressions_extended.parquet
    metrics:
      - data/gtex/output/all/gtex_metrics_best_huber.json
      - data/gtex/output/all/gtex_metrics_best_mse.json
    plots:
      - data/gtex/output/all/gtex_metrics_plot.json:
          y: "MAE"
          x_label: steps
          y_label: years
          title: "aging clock trained on all samples"