Snakemake: do not resolve wildcards in input

I want to use a pipeline that requires an * and R{1,2} in the input string.

rule preprocess:
    input:
        reads = '{WORKDIR}/*/*R{1,2}.fastq.gz'
    output:
        bam = '{WORKDIR}/mapping/bwa/{sample}.bam'
    shell:
        "nextflow run somePipeline "
        "--reads '{input.reads}' "

The final shell command should be: nextflow run somePipeline --reads 'work_dir/main_sample_folder/*/*R{1,2}.fastq.gz'

However, Snakemake thinks 'R{1,2} is a wildcard and produces the following error:

Wildcards in input files cannot be determined from output files:
'1'

I've tried escaping the curly brackets with double curly brackets R{{1,2}} and with backslash R\{1,2\}, but it did not work.

How to escape R{1,2} part of the input, so that Snakemake does not think this is a wildcard, rather a part of the string?