######################
## Global variables ##
######################

samples: ["E7.5_rep1", "E7.5_rep2", "E7.75_rep1", "E8.0_rep1", "E8.0_rep2", "E8.5_rep1", "E8.5_rep2", "E8.75_rep1", "E8.75_rep2", "E8.5_CRISPR_T_KO", "E8.5_CRISPR_T_WT"]
reference_samples: ["E7.5_rep1", "E7.5_rep2", "E7.75_rep1", "E8.0_rep1", "E8.0_rep2", "E8.5_rep1", "E8.5_rep2", "E8.75_rep1", "E8.75_rep2"]
wt_ko_samples: ["E8.5_CRISPR_T_KO", "E8.5_CRISPR_T_WT"]
celltypes: ["Epiblast", "Primitive_Streak", "Caudal_epiblast", "PGC", "Anterior_Primitive_Streak", "Notochord", "Def._endoderm", "Gut", "Nascent_mesoderm", "Mixed_mesoderm", "Intermediate_mesoderm", "Caudal_Mesoderm", "Paraxial_mesoderm", "Somitic_mesoderm", "Pharyngeal_mesoderm", "Cardiomyocytes", "Allantois", "ExE_mesoderm", "Mesenchyme", "Haematoendothelial_progenitors", "Endothelium", "Blood_progenitors_1", "Blood_progenitors_2", "Erythroid1", "Erythroid2", "Erythroid3", "NMP", "Rostral_neurectoderm", "Caudal_neurectoderm", "Neural_crest", "Forebrain_Midbrain_Hindbrain", "Spinal_cord", "Surface_ectoderm", "Visceral_endoderm", "ExE_endoderm", "ExE_ectoderm", "Parietal_endoderm"]
stages: ["E7.5","E7.75","E8.0","E8.5","E8.75"]
trajectories: ["nmp"]

#################
## Directories ##
#################

directories:
    base: /bi/group/reik/ricard/data/gastrulation_multiome_10x
    original_data: /bi/group/reik/ricard/data/gastrulation_multiome_10x/original
    processed_data: /bi/group/reik/ricard/data/gastrulation_multiome_10x/test/processed/rna
    results: /bi/group/reik/ricard/data/gastrulation_multiome_10x/test/results
    # processed_data: /bi/group/reik/ricard/data/gastrulation_multiome_10x/processed/rna
    # results: /bi/group/reik/ricard/data/gastrulation_multiome_10x/results
    atlas: /bi/group/reik/ricard/data/pijuansala2019_gastrulation10x

###############
## Resources ##
###############

resources:
    python: /bi/group/reik/ricard/software/miniconda3/envs/main/bin/python
    # marker_genes: /bi/group/reik/ricard/data/pijuansala2019_gastrulation10x/results/marker_genes/all_stages/marker_genes.txt.gz
    # TFs_file: "/bi/group/reik/ricard/data/mm10_regulation/TFs/TFs.txt"
    TFs_file: "/bi/group/reik/ricard/data/mm10_regulation/TFs/TFs.txt"
    cell_metadata: "/bi/group/reik/ricard/data/gastrulation_multiome_10x/results/atac/archR/celltype_assignment/sample_metadata_after_celltype_assignment.txt.gz"
    mm10_mask: /bi/group/reik/ricard/data/mm10_sequence/repeats/mm10_rmsk.gtf
    genes_gtf: /bi/scratch/Stephen_Clark/annotations/gtf/Mus_musculus.GRCm38.98.gtf

#############
## Scripts ##
#############

scripts:
    # basic processing
    create_seurat: ../processing/1_create_seurat_rna.R
    qc: ../processing/2_QC.R
    seurat_to_sce: ../processing/3_seurat_to_SCE.R
    doublet_detection: ../processing/4_doublet_detection.R
    parse_doublets: ../processing/5_parse_sample_metadata_after_doublets.R
    plot_stats_per_sample: ../processing/6_plot_stats.R
    convert_SingleCellExperiment_to_anndata: ../conversions/convert_SingleCellExperiment_to_anndata.R

    # mapping
    mapping_mnn: ../mapping/run/mnn/mapping_mnn.R
    mapping_seurat: ../mapping/run/seurat/mapping_seurat.R
    parse_mapping: ../mapping/run/parse_sample_metadata_after_mapping.R

    # dimensionality reduction
    dimensionality_reduction_sce: ../dimensionality_reduction/dimensionality_reduction_sce.R
    dimensionality_reduction_seurat: ../dimensionality_reduction/dimensionality_reduction_seurat.R

    # pseudobulk
    pseudobulk_rna: ../pseudobulk/pseudobulk_rna.R
    pseudobulk_rna_with_replicates: ../pseudobulk/pseudobulk_rna_with_replicates.R

    # cell type validation
    celltype_validation_single_cells: ../celltype_validation/celltype_validation_single_cells.R
    celltype_validation_pseudobulk: ../celltype_validation/celltype_validation_pseudobulk.R

    # mapping
    plot_mapping_results: ../mapping/analysis/plot_mapping_umap.R
    plot_celltype_proportions: ../celltype_proportions/plot_celltype_proportions.R

    # TF2gene coexpression
    infer_trajectories: ../trajectories/infer_trajectory.R
    coexpression_TF_vs_gene_single_cells: ../TF2gene_coexpression/coexpression_TF_vs_gene_single_cells.R
    coexpression_TF_vs_gene_pseudobulk: ../TF2gene_coexpression/coexpression_TF_vs_gene_pseudobulk.R

    # metacells
    run_metacells: ../metacells/run/run_metacell.py
    run_metacells_trajectory: ../metacells/run/run_metacell_trajectory.py
    aggregate_rna_metacells: ../metacells/run/aggregate_rna_metacell.R
    coexpression_TF_vs_gene_metacells: ../TF2gene_coexpression/coexpression_TF_vs_gene_metacells.R

    # differential expression between cell types
    differential_celltype_cells: ../differential/cells/differential.R
    differential_celltype_metacells: ../differential/metacells/differential.R
    differential_celltype_pseudobulk: ../differential/pseudobulk/celltype/differential_celltype_pseudobulk.R

    # differential expression between genotypes
    # differential_genotype_cells: ../differential/cells/genotype/run_diff_expr_genotype.R
    # differential_genotype_metacells: ../differential/metacells/genotype/run_diff_expr_genotype.R
    differential_celltype_genotype_pseudobulk: ../differential/pseudobulk/celltype_genotype/differential_celltype_genotype_pseudobulk.R

    # parse differential expression results
    parse_differential_celltype_cells: ../differential/cells/parse_differential_results.R
    parse_differential_celltype_metacells: ../differential/metacells/parse_differential_results.R
    parse_differential_celltype_pseudobulk: ../differential/pseudobulk/celltype/parse_differential_results.R
    parse_differential_celltype_genotype_pseudobulk: ../differential/pseudobulk/celltype_genotype/parse_differential_results.R


    # extract TFs from differential results
    differential_extract_TFs: ../differential/other/extract_TFs_diff.R

    # Marker genes
    define_marker_genes_cells: ../differential/cells/celltype/analysis/define_marker_genes.R
    define_marker_genes_metacells: ../differential/metacells/celltype/analysis/define_marker_genes.R
    define_marker_genes_pseudobulk: ../differential/pseudobulk/celltype/analysis/define_marker_genes.R
    define_marker_TFs_cells: ../differential/cells/celltype/analysis/TFs/define_marker_TFs.R
    define_marker_TFs_metacells: ../differential/metacells/celltype/analysis/TFs/define_marker_TFs.R
    define_marker_TFs_pseudobulk: ../differential/pseudobulk/celltype/analysis/define_marker_TFs.R

    # Extract TFs
    extract_TF_SingleCellExperiment: ../processing/extract_TFs_from_SingleCellExperiment.R

    # Velocity
    create_anndata_velocyto: ../scanpy/velocyto/create_anndata_from_loom_files.py


################
## Parameters ##
################

pseudobulk_rna:
    # group_by: ["sample", "celltype_mnn", "celltype_seurat"]
    group_by: ["celltype", "sample", "celltype_genotype"]
    normalisation_method: "cpm"
    
pseudobulk_rna_with_replicates:
    nrep: 5
    min_cells: 25
    fraction_cells_per_replicate: 0.30

doublet_detection:
    doublet_score_threshold: 1.25

mapping_mnn:
    atlas_stages: ["E6.5","E6.75","E7.0","E7.25","E7.5","E7.75","E8.0","E8.25","E8.5"]
    npcs: 50
    n_neighbours: 25

mapping_seurat:
    atlas_stages: ["E6.5","E6.75","E7.0","E7.25","E7.5","E7.75","E8.0","E8.25","E8.5"]
    npcs: 50
    n_neighbours: 25

qc:
    min_nFeature_RNA: 2000
    max_nFeature_RNA: 10000
    percent_mt: 40
    percent_rib: 20

dimensionality_reduction_sce:
    npcs: 50
    features: 2500
    n_neighbors: 25
    min_dist: 0.5
    vars_to_regress: ["nFeature_RNA"]
    batch_variable: ["sample","None"]
    remove_ExE_cells: ["True","False"]
    # colour_by: ["sample", "stage", "celltype_mnn", "celltype_seurat", "doublet_score", "doublet_call", "nFeature_RNA"]
    colour_by: ["sample", "stage", "celltype", "nFeature_RNA"]

dimensionality_reduction_seurat:
    npcs: 50
    features: 2500
    n_neighbors: 25
    min_dist: 0.5
    seed: 42
    vars_to_regress: ["nFeature_RNA"]
    batch_variable: ["sample"]
    remove_ExE_cells: ["True","False"]
    # vars_to_regress: ["nFeature_RNA","mitochondrial_percent_RNA"]
    colour_by: ["sample", "stage", "celltype", "nFeature_RNA"]

celltype_validation_single_cells:
    celltype_label: "celltype"

celltype_validation_pseudobulk:
    celltype_label: "celltype"

plot_celltype_proportions:
    celltype_label: "celltype"

infer_trajectories:
    # trajectory_name: ["blood", "ectoderm", "endoderm", "mesoderm"]
    trajectory_name: ["blood"]
    celltype_label: "celltype"

# coexpression_TF_vs_gene_trajectories

run_metacells:
    percent_metacells: 0.05
    n_features: 1500
    n_pcs: 25

run_metacells_trajectory:
    percent_metacells: 0.05
    n_features: 1500
    n_pcs: 15
    trajectories: ["nmp"]

aggregate_rna_metacells:
    metacell_min_reads: 25000
    normalisation_method: "cpm"

differential_celltype_cells:
    group_variable: ["celltype"]
    min_cells: 25

differential_celltype_metacells:
    group_variable: ["celltype"]
    min_cells: 5

differential_genotype_metacells:
    group_variable: ["genotype"]
    min_cells: 5

differential_genotype_pseudobulk:
    min_cells: 50

differential_genotype:
    group_variable: ["genotype"]
    ko_class: "T_KO"
    wt_class: "WT"
    min_cells: 50

tf2gene_cor:
    cor_test: ["pearson","spearman"]

define_marker_genes:
    min_fold_change: 1.0
    min_score: 0.85
    fdr: 0.01

define_marker_TFs:
    min_fold_change: 1.0
    min_score: 0.85
    fdr: 0.01

###########
## Slurm ##
###########

slurm:  # memory in MB
    create_seurat: 
        memory: 50000
        threads: 1
    qc: 
        memory: 3000
        threads: 1
    seurat_to_sce:
        memory: 50000
        threads: 1
    doublet_detection:
        memory: 25000
        threads: 1
    parse_doublet_results:
        memory: 2000
        threads: 1
    plot_stats_per_sample:
        threads: 1
        memory: 15000
    pseudobulk_rna:
        memory: 15000
        threads: 1
    pseudobulk_rna_with_replicates:
        memory: 25000
        threads: 1
    mapping_mnn:
        memory: 50000
        threads: 1
    mapping_mnn_all_samples:
        memory: 50000
        threads: 1
    mapping_seurat:
        memory: 150000
        threads: 1
    mapping_seurat_all_samples:
        memory: 150000
        threads: 1
    parse_mapping_results:
        memory: 2000
        threads: 1
    dimensionality_reduction_sce:
        memory: 20000
        threads: 1
    dimensionality_reduction_seurat:
        memory: 70000
        threads: 1
    dimensionality_reduction_seurat_per_stage:
        memory: 30000
        threads: 1
    celltype_validation_single_cells:
        memory: 15000
        threads: 1
    celltype_validation_pseudobulk:
        memory: 3000
        threads: 1
    plot_mapping_results:
        memory: 3000
        threads: 1
    plot_celltype_proportions:
        memory: 3000
        threads: 1
    infer_trajectories:
        memory: 10000
        threads: 1
    coexpression_TF_vs_gene_single_cells:
        memory: 80000
        threads: 1
    coexpression_TF_vs_gene_single_cells_denoised:
        memory: 220000
        threads: 1
    coexpression_TF_vs_gene_pseudobulk:
        memory: 5000
        threads: 1
    convert_SingleCellExperiment_to_anndata:
        memory: 45000
        threads: 1
    run_metacells:
        threads: 1
        memory: 25000
    run_metacells_trajectory:
        threads: 1
        memory: 15000
    aggregate_rna_metacells:
        threads: 1
        memory: 15000
    coexpression_TF_vs_gene_metacells:
        threads: 1
        memory: 15000
    samtools_sort:
        threads: 8
        memory: 50000
    run_velocyto:
        threads: 8
        memory: 50000
    differential_cells:
        threads: 1
        memory: 10000
    differential_metacells:
        threads: 1
        memory: 5000
    differential_pseudobulk:
        threads: 1
        memory: 3000
    parse_differential_celltype:
        threads: 1
        memory: 15000
    differential_celltype_pseudobulk:
        threads: 1
        memory: 7000
    differential_genotype_pseudobulk:
        threads: 1
        memory: 7000
    differential_celltype_extract_TFs:
        threads: 1
        memory: 7000
    extract_TF_SingleCellExperiment_cells:
        threads: 1
        memory: 20000
    extract_TF_SingleCellExperiment_metacells:
        threads: 1
        memory: 15000
    extract_TF_SingleCellExperiment_pseudobulk:
        threads: 1
        memory: 8000
    define_marker_genes:
        threads: 1
        memory: 10000
    define_marker_TFs:
        threads: 1
        memory: 5000
    create_anndata_velocyto:
        threads: 1
        memory: 30000