Paths from json file don't expand in Snakemake

Question

I have a Snakemake pipeline where I get my input/output paths for my file folders from a json file and use the expand function to obtain the paths.

import json

with open('config.json', 'r') as f:
    config = json.load(f)

wildcard = ["1234", "5678"]

rule them_all:
    input:
        expand('config["data_input"]/data_{wc}.tab', wc = wildcard)

    output:
        expand('config["data_output"]/output_{wc}.rda', wc = wildcard)
    shell:
        "Rscript ./my_script.R"

My config.json is

{
"data_input": "/very/long/path",
"data_output": "/slightly/different/long/path"
}

While trying to make a dry run, though, I get the following error:

$ snakemake -np
Building DAG of jobs...
MissingInputException in line 12 of /path/to/Snakefile:
Missing input files for rule them_all:
config["data_input"]/data_1234.tab
config["data_input"]/data_5678.tab

The files are there and their path is /very/long/path/data_1234.tab.

This is probably a low-hanging fruit, but what am I doing wrong in the syntax for the expansion? Or is it the way I call the json file?

Answer 1

expand() does not interpret access to dictionaries for its first argument while expanding the path with quotation marks, so this operation with expand() has to be done in a wildcard. The correct syntax, in this case, would be e.g.

expand('{input_folder}/data_{wc}.tab', wc = wildcard, input_folder = config["data_input"])