Snakefile_multiome

from snakemake.utils import R
import glob
import os
import itertools

include: "rules/singlecell_import.smk"
include: "rules/singlecell_multiome_import.smk"

## if there is only one sample, there is no need to run cellrange aggr 
aggregate = getattr(config, "aggregate", True)
aggr_output4ruleall = ["aggregate.complete"] if aggregate else []

## only one sample or Multiplexing Capture 
if len(samples) == 1 or hasattr(config, "cmo"):
    aggregate = False
    aggr_output4ruleall = []

include: "rules/singlecell_rules_common.smk"

copy_complete = 'copy.complete'

if external == True:
    wreport_result = []
    xreport_result = []
    copy_complete = []

rule all:
    input: 
        expand("run_{sample}_10x_cellranger_count.err", sample=samples), 
        "finalreport/metric_summary.xlsx", 
        expand("finalreport/summaries/{sample}_web_summary.html", sample=samples), 
        aggr_output4ruleall, 
        xreport_result, 
        wreport_result, 
        copy_complete,  
        rule_all_append

rule librariesCSV:
    output: expand("{sample}_libraries.csv", sample=samples)
    params: fastq = ",".join(config.unaligned)
    shell: "python {analysis}/workflow/scripts/fb/create_library_files.py {config.libraries} {params.fastq}"

include_introns = getattr(config, "include_introns", True) 
def count_premrna(wildcards):
    if include_introns is False:
        return('--gex-exclude-introns')
    else:
        return(' ')

rule arc_count:
    input: "{sample}_libraries.csv"
    output: "{sample}/outs/web_summary.html"
    log: err = "run_{sample}_10x_cellranger_count.err", log ="run_{sample}_10x_cellranger_count.log"
    params: prefix = "{sample}", prefix2 = filterFastq,count_intronic = count_premrna
    container: program.cellranger_arc
    shell:
        """
rm -r {params.prefix}; cellranger-arc count {params.count_intronic} --id={params.prefix} --reference={reference.arc_reference} --libraries={input} 2>{log.err} 1>{log.log}
"""

rule summaryFiles:
    input: expand("{sample}/outs/web_summary.html", sample=samples)
    output: "finalreport/metric_summary.xlsx", expand("finalreport/summaries/{sample}_web_summary.html", sample=samples)
    shell: "python {analysis}/workflow/scripts/multiome/generateSummaryFiles.py"

if aggregate == True:
    rule aggregateCSV:
        input: expand("{sample}/outs/web_summary.html", sample=samples)
        output: "AggregatedDatasets.csv"
        shell: "python {analysis}/workflow/scripts/multiome/generateAggregateCSV.py {analysis}"

    rule aggregate:
        input: csv="AggregatedDatasets.csv"
        output: touch('aggregate.complete')
        log: err="run_10x_aggregate.err", log="run_10x_aggregate.log"
        container: program.cellranger_arc
        shell: "cellranger-arc aggr --id=AggregatedDatasets --csv={input.csv}  --reference={reference.arc_reference} --normalize=depth 2>{log.err} 1>{log.log}"