|
| 1 | +#################################################################################################### |
| 2 | +# Compares DivRef 1.1 single-variant sites against allele frequencies from different gnomAD |
| 3 | +# releases for chr22. |
| 4 | +#################################################################################################### |
| 5 | + |
| 6 | +from pathlib import Path |
| 7 | + |
| 8 | +#################################################################################################### |
| 9 | +# Inputs / constants |
| 10 | +#################################################################################################### |
| 11 | + |
| 12 | +OUTPUT_DIR: Path = Path("data/analysis") |
| 13 | +CONTIG: str = "chr22" |
| 14 | +DIVREF_DUCKDB_URL: str = ( |
| 15 | + "https://zenodo.org/records/14802613/files/" "DivRef-v1.1.haplotypes_gnomad_merge.index.duckdb" |
| 16 | +) |
| 17 | +GNOMAD_VERSIONS: list[str] = ["joint_41", "hgdp_1kg_312"] |
| 18 | + |
| 19 | +# Maps filename wildcard → plot label used by compare_divref_gnomad.R |
| 20 | +GNOMAD_LABEL: dict[str, str] = { |
| 21 | + "joint_41": "gnomAD 4.1 joint AF", |
| 22 | + "hgdp_1kg_312": "gnomAD 3.1.2 HGDP+1KG AF", |
| 23 | +} |
| 24 | + |
| 25 | +#################################################################################################### |
| 26 | +# Rules |
| 27 | +#################################################################################################### |
| 28 | + |
| 29 | +ruleorder: compare_divref_gnomad > extract_gnomad_single_afs |
| 30 | + |
| 31 | +rule all: |
| 32 | + input: |
| 33 | + expand( |
| 34 | + f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.af_diffs.png", |
| 35 | + gnomad_version=GNOMAD_VERSIONS, |
| 36 | + ), |
| 37 | + expand( |
| 38 | + f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.not_in_gnomad_afs.png", |
| 39 | + gnomad_version=GNOMAD_VERSIONS, |
| 40 | + ), |
| 41 | + expand( |
| 42 | + f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.divref_not_in_gnomad.tsv", |
| 43 | + gnomad_version=GNOMAD_VERSIONS, |
| 44 | + ), |
| 45 | + expand( |
| 46 | + f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.log", |
| 47 | + gnomad_version=GNOMAD_VERSIONS, |
| 48 | + ) |
| 49 | + |
| 50 | + |
| 51 | +#################################################################################################### |
| 52 | +# Downloads the DivRef 1.1 DuckDB index from Zenodo. |
| 53 | +#################################################################################################### |
| 54 | +rule download_divref_index: |
| 55 | + output: |
| 56 | + duckdb=f"{OUTPUT_DIR}/input/DivRef-v1.1.haplotypes_gnomad_merge.index.duckdb", |
| 57 | + log: |
| 58 | + f"logs/compare_divref_gnomad/download_divref_index.log", |
| 59 | + params: |
| 60 | + url=DIVREF_DUCKDB_URL, |
| 61 | + shell: |
| 62 | + """ |
| 63 | + ( |
| 64 | + wget --no-verbose -O {output.duckdb} {params.url} |
| 65 | + ) &> {log} |
| 66 | + """ |
| 67 | + |
| 68 | + |
| 69 | +#################################################################################################### |
| 70 | +# Extracts chr22 allele frequencies from a gnomAD sites table and writes a Hail table and a flat |
| 71 | +# TSV with one AF column per population. |
| 72 | +#################################################################################################### |
| 73 | +rule extract_gnomad_single_afs: |
| 74 | + output: |
| 75 | + ht=directory(f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.ht"), |
| 76 | + tsv=f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.tsv", |
| 77 | + log: |
| 78 | + f"logs/compare_divref_gnomad/extract_gnomad_single_afs.{{gnomad_version}}.log", |
| 79 | + params: |
| 80 | + contig=CONTIG, |
| 81 | + gnomad_version=lambda wildcards: wildcards.gnomad_version.upper(), |
| 82 | + shell: |
| 83 | + """ |
| 84 | + ( |
| 85 | + divref extract-gnomad-single-afs \ |
| 86 | + --contig {params.contig} \ |
| 87 | + --gnomad-version {params.gnomad_version} \ |
| 88 | + --out-sites-hail-table {output.ht} \ |
| 89 | + --out-sites-tsv {output.tsv} |
| 90 | + ) &> {log} |
| 91 | + """ |
| 92 | + |
| 93 | + |
| 94 | +#################################################################################################### |
| 95 | +# Compares DivRef 1.1 gnomAD_variant sites to the extracted gnomAD allele frequencies. |
| 96 | +# If all DivRef variants are found in gnomAD the R script exits early; touch ensures Snakemake's |
| 97 | +# output checks are satisfied even in that case. |
| 98 | +#################################################################################################### |
| 99 | +rule compare_divref_gnomad: |
| 100 | + input: |
| 101 | + duckdb=f"{OUTPUT_DIR}/input/DivRef-v1.1.haplotypes_gnomad_merge.index.duckdb", |
| 102 | + tsv=f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.tsv", |
| 103 | + output: |
| 104 | + af_diffs_png=f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.af_diffs.png", |
| 105 | + not_in_gnomad_png=f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.not_in_gnomad_afs.png", |
| 106 | + not_in_gnomad_tsv=f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.divref_not_in_gnomad.tsv", |
| 107 | + log: |
| 108 | + f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}.log", |
| 109 | + params: |
| 110 | + contig=CONTIG, |
| 111 | + gnomad_label=lambda wildcards: GNOMAD_LABEL[wildcards.gnomad_version], |
| 112 | + output_base=f"{OUTPUT_DIR}/compare_divref_gnomad/{CONTIG}.{{gnomad_version}}", |
| 113 | + shell: |
| 114 | + """ |
| 115 | + ( |
| 116 | + Rscript scripts/compare_divref_gnomad.R \ |
| 117 | + --contig {params.contig} \ |
| 118 | + --divref_duckdb {input.duckdb} \ |
| 119 | + --gnomad_tsv {input.tsv} \ |
| 120 | + --gnomad_label '{params.gnomad_label}' \ |
| 121 | + --output_base {params.output_base} |
| 122 | +
|
| 123 | + # The R script exits early when no DivRef variants are absent from gnomAD, |
| 124 | + # so touch ensures these outputs exist for Snakemake's file checks. |
| 125 | + touch {output.not_in_gnomad_png} {output.not_in_gnomad_tsv} |
| 126 | + ) &> {log} |
| 127 | + """ |
0 commit comments