workflow/Snakefile

# =====================================================
# YuanSeq Snakemake Workflow
# 转录组差异分析与功能富集 - 可复现命令行流程
# 对应 YuanSeq Shiny 应用的逻辑架构
# =====================================================

configfile: "config/config.yaml"

# 路径变量
OUTPUT_DIR = config["output"]["dir"]
PREFIX = config["output"]["prefix"]
SCRIPT_DIR = "scripts"

# 输入文件
INPUT_COUNTS = config["input"]["counts"]
INPUT_DESIGN = config["input"]["design"]

# 输出文件路径
DEG_FILE = f"{OUTPUT_DIR}/{PREFIX}_deg_results.csv"
VOLCANO_PDF = f"{OUTPUT_DIR}/{PREFIX}_volcano.pdf"
KEGG_CSV = f"{OUTPUT_DIR}/{PREFIX}_kegg_enrichment.csv"
KEGG_PDF = f"{OUTPUT_DIR}/{PREFIX}_kegg_dotplot.pdf"
GO_CSV = f"{OUTPUT_DIR}/{PREFIX}_go_enrichment.csv"
GO_PDF = f"{OUTPUT_DIR}/{PREFIX}_go_dotplot.pdf"
GSEA_CSV = f"{OUTPUT_DIR}/{PREFIX}_gsea_results.csv"
GSEA_PDF = f"{OUTPUT_DIR}/{PREFIX}_gsea_ridge.pdf"
TF_CSV = f"{OUTPUT_DIR}/{PREFIX}_tf_activity.csv"
TF_PDF = f"{OUTPUT_DIR}/{PREFIX}_tf_activity.pdf"

# =====================================================
# Rule: 差异分析 (limma-voom / edgeR)
# =====================================================
rule differential_analysis:
    input:
        counts = INPUT_COUNTS,
        design = INPUT_DESIGN
    output:
        deg = DEG_FILE
    params:
        species = config["species"],
        pval_cutoff = config["differential"]["pval_cutoff"],
        log2fc_cutoff = config["differential"]["log2fc_cutoff"],
        method = config["differential"].get("method", "auto")
    log:
        f"{OUTPUT_DIR}/logs/differential_analysis.log"
    shell:
        "mkdir -p {OUTPUT_DIR}/logs && "
        "Rscript {SCRIPT_DIR}/01_differential_analysis.R "
        "--counts {input.counts} --design {input.design} "
        "--output {output.deg} "
        "--species {params.species} "
        "--pval_cutoff {params.pval_cutoff} "
        "--log2fc_cutoff {params.log2fc_cutoff} "
        "--method {params.method} "
        "2>&1 | tee {log}"


# =====================================================
# Rule: 火山图
# =====================================================
rule volcano_plot:
    input:
        deg = DEG_FILE
    output:
        pdf = VOLCANO_PDF
    params:
        pval_cutoff = config["differential"]["pval_cutoff"],
        log2fc_cutoff = config["differential"]["log2fc_cutoff"]
    log:
        f"{OUTPUT_DIR}/logs/volcano.log"
    shell:
        "Rscript {SCRIPT_DIR}/02_volcano_plot.R "
        "--deg {input.deg} --output {output.pdf} "
        "--pval_cutoff {params.pval_cutoff} "
        "--log2fc_cutoff {params.log2fc_cutoff} "
        "2>&1 | tee {log}"


# =====================================================
# Rule: KEGG 富集
# =====================================================
rule kegg_enrichment:
    input:
        deg = DEG_FILE
    output:
        csv = KEGG_CSV,
        pdf = KEGG_PDF
    params:
        species = config["kegg_species"],
        direction = config["kegg"]["direction"],
        pval_cutoff = config["kegg"]["pvalueCutoff"]
    log:
        f"{OUTPUT_DIR}/logs/kegg.log"
    shell:
        "Rscript {SCRIPT_DIR}/03_kegg_enrichment.R "
        "--deg {input.deg} --output_csv {output.csv} --output_pdf {output.pdf} "
        "--species {params.species} --direction {params.direction} "
        "--pval_cutoff {params.pval_cutoff} "
        "2>&1 | tee {log}"


# =====================================================
# Rule: GO 富集
# =====================================================
rule go_enrichment:
    input:
        deg = DEG_FILE
    output:
        csv = GO_CSV,
        pdf = GO_PDF
    params:
        species = config["species"],
        ontology = config["go"]["ontology"],
        pval_cutoff = config["go"]["pvalueCutoff"]
    log:
        f"{OUTPUT_DIR}/logs/go.log"
    shell:
        "Rscript {SCRIPT_DIR}/04_go_enrichment.R "
        "--deg {input.deg} --output_csv {output.csv} --output_pdf {output.pdf} "
        "--species {params.species} --ontology {params.ontology} "
        "--pval_cutoff {params.pval_cutoff} "
        "2>&1 | tee {log}"


# =====================================================
# Rule: GSEA (需在 config 中设置 gsea.gmt_file)
# 运行: snakemake results/yuanseq_gsea_results.csv
# =====================================================
rule gsea_analysis:
    input:
        deg = DEG_FILE,
        gmt = config["gsea"]["gmt_file"]
    output:
        csv = GSEA_CSV,
        pdf = GSEA_PDF
    params:
        pval_cutoff = config["gsea"]["pvalueCutoff"],
        id_type = config["gsea"]["id_type"]
    log:
        f"{OUTPUT_DIR}/logs/gsea.log"
    shell:
        "Rscript {SCRIPT_DIR}/05_gsea_analysis.R "
        "--deg {input.deg} --gmt {input.gmt} "
        "--output_csv {output.csv} --output_pdf {output.pdf} "
        "--pval_cutoff {params.pval_cutoff} --id_type {params.id_type} "
        "2>&1 | tee {log}"


# =====================================================
# Rule: 转录因子活性
# =====================================================
rule tf_activity:
    input:
        deg = DEG_FILE
    output:
        csv = TF_CSV,
        pdf = TF_PDF
    params:
        species = config["species"],
        method = config["tf_activity"].get("method", "ulm"),
        collectri_dir = config["tf_activity"].get("collectri_dir", ".")
    log:
        f"{OUTPUT_DIR}/logs/tf_activity.log"
    shell:
        "Rscript {SCRIPT_DIR}/06_tf_activity.R "
        "--deg {input.deg} --output_csv {output.csv} --output_pdf {output.pdf} "
        "--species {params.species} --method {params.method} "
        "--collectri_dir {params.collectri_dir} "
        "2>&1 | tee {log}"


# =====================================================
# 主目标：核心流程 (DEG -> 火山图 -> KEGG -> GO -> TF)
# =====================================================
rule all:
    input:
        DEG_FILE,
        VOLCANO_PDF,
        KEGG_CSV,
        KEGG_PDF,
        GO_CSV,
        GO_PDF,
        TF_CSV,
        TF_PDF