config.json

{
    "last_step": "STARsolo_kb_solo",
    "wet_lab_info": "/sc/arion/projects/psychAD/upload2synpase/final_NPSAD_snRNAseq_metrics_combo.csv",
    "columns_to_pick": ["cDNA_ID", "hashtag", "SubID"],
    "select_fastqs": "/sc/arion/projects/psychAD/snakemake_pipeline/fastq_files.txt",
    "cDNA_fastqs_dir": "/sc/arion/projects/psychAD/fastq_test_run/",
    "R1_suffix": ".R1.fastq.gz",
    "R2_suffix": ".R2.fastq.gz",
    "HTO_fastqs_dir": "/sc/arion/projects/psychAD/fastq_test_run/",
    
    "bams_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/STARsolo_bams/",
    "sjdboverhang": 89,
    "genome_dir": "/sc/arion/projects/psychAD/star_275b_gen_89o",
    "gtf_file": "/sc/arion/projects/psychAD/ref_GRCh38/Homo_sapiens.GRCh38.104.gtf",
    "STAR_version": "star/2.7.8a",
    "solo_cell_filter": "EmptyDrops_CR",
    "soloType": "CB_UMI_Simple",
    "umi_len": 12,
    "whitelist": "/sc/arion/projects/psychAD/10x_vs_barcodes/3M-february-2018.txt",
    "star_params_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/dual_test/STARsolo_params/",
    "SAM_attr": "NH HI AS nM NM MD CB UB CR UR GX GN sS sQ sM",
    "features": "Gene GeneFull SJ Velocyto",
    "bam": "_Aligned.sortedByCoord.out.bam",
    "gene_raw_matrix": "_Solo.out/Gene/raw/matrix.mtx",
    "genefull_raw_matrix": "_Solo.out/GeneFull/raw/matrix.mtx",
    "STAR_log_final": "_Log.final.out",
    "gene_features": "_Solo.out/Gene/Features.stats",
    "gene_summary": "_Solo.out/Gene/Summary.csv",
    "genefull_features": "_Solo.out/GeneFull/Features.stats",
    "genefull_summary": "_Solo.out/GeneFull/Summary.csv",
    "barcodes_stats": "_Solo.out/Barcodes.stats",
    "bai": "_Aligned.sortedByCoord.out.bam.bai",
    "gene_lun_matrix": "_Solo.out/Gene/filtered/matrix.mtx.gz",
    "gene_lun_features": "_Solo.out/Gene/filtered/features.tsv.gz",
    "gene_lun_barcodes": "_Solo.out/Gene/filtered/barcodes.tsv.gz",
    "genefull_lun_matrix": "_Solo.out/GeneFull/filtered/matrix.mtx.gz",
    "genefull_lun_features": "_Solo.out/GeneFull/filtered/features.tsv.gz",
    "genefull_lun_barcodes": "_Solo.out/GeneFull/filtered/barcodes.tsv.gz",


    "genome_fasta": "/sc/arion/projects/psychAD/ref_GRCh38/Homo_sapiens.GRCh38.dna.primary_assembly.fa",
    "window_size": 100,
    "gc_bias_metrics": "_gc_bias_metrics.txt",
    "gc_summary_metrics": "_summary_metrics.txt",

    "flat_ref": "/sc/arion/projects/psychAD/ref_GRCh38/refFlat_wo_chr.txt",
    "strand": "NONE",
    "rnaseq_metrics": "_rnaseq_metrics.txt",

    "kallisto_bustools_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/HTO_info/feature_barcodes/",
    "featuremap_script": "/sc/arion/projects/psychAD/pnm/kite/featuremap/featuremap.py",
    "feature_barcodes": "FeatureBarcodes.csv",
    "headers": "yes",
    "features_mismatch_fa": "FeaturesMismatch.fa",
    "features_mismatch_t2g": "FeaturesMismatch.t2g",
    "k_mer_length": 15,
    "chemistry": "10xv3",
    "features_mismatch_idx": "FeaturesMismatch.idx",
    "bus_file": "output.bus",
    "ec_matrix": "matrix.ec",
    "tx": "transcripts.txt",
    "bus_file_corrected": "output_corrected.bus",
    "bus_file_sorted": "output_sorted.bus",
    "bus_count_dir": "bus_count_out/",
    "bus_count_mtx": "output.mtx",
    "bus_count_genes": "output.genes.txt",
    "bus_count_barcodes": "output.barcodes.txt",
    "h5ad_bustools_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/cache/",

    "calico_solo_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/demultiplex/solo/",
    "calico_solo_h5ad": "_STARsolo_out.h5ad",
    "max_mito_percentage": 5,
    "min_genes_per_cell": 1000,
    "min_cells_per_gene": 10,
    "final_count_matrix_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/final_count_matrix/solo/",
    "final_count_matrix_h5ad": "_STARsolo_out.h5ad",
    "demultiplex_info_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/demultiplex/info/",
    "demultiplex_info": "_STARsolo_info.tsv",
    "gene_info_file": "/sc/arion/projects/psychAD/pnm/Hs_allchr_MT.txt",

    "cellsnp_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/demux_gt/cellSNP_imp/",
    "filt_barcodes_dir": "/sc/arion/projects/psychAD/snakemake_pipeline/inp_for_vireo/",
    "filt_barcodes": "filt_bc_vireo_", 
    "cellsnp_cells": "cellSNP.cells.vcf.gz",
    "cellsnp_base": "cellSNP.base.vcf.gz",
    "ref_snps": "/sc/arion/projects/psychAD/Single_cell_data/cellSNP_ref/CMC_SNParray_imputed_MAFp05_cellSNP_ref.vcf",
    "umi_tag": "UB",
    "cell_tag": "CB",
    "n_proc": 20,
    "min_maf": 0.1,
    "min_aggr_count": 20,

    "vireosnp_dir": "/sc/arion/projects/psychAD/demux_gt/vireoSNP_imp/",
    "donors_vcf": "GT_donors.vireo.vcf.gz",
    "filt_vcf_dir": "/sc/arion/projects/psychAD/demux_gt/bcf_filt_vcf_imp/",
    "filt_vcf": "filt.vcf.gz",
    "genotyped_vcf": "/sc/arion/projects/psychAD/pnm/CMC_MSSM_NPSAD_genotyped_all.vcf.gz",
    "meta_data_geno_samp": "/sc/arion/projects/psychAD/pnm/Hash_map_gtID_SubID.csv",
    "donor_genotype": "GT",
    
    "split_bams_dir": "/sc/arion/projects/psychAD/STARsolo_split_filt_bams/",
    "split_bams_proxy_dir": "/sc/arion/projects/psychAD/pnm/split_bam_logs/",

    "meta_data": "/sc/arion/projects/psychAD/pnm/Final_out_MAP_2.tsv",
    "log_all_stats": "/sc/arion/projects/psychAD/pnm/All_logs.tsv",
    "meta_data_headers": ["val_in_log", "curr_val", "prog", "sub_prog", "desc"]

}