I am not getting why snakemake is not able to identify the dependencies in this simple example. I am expecting that it identify that the input of main is generated in simulation and the inputs of the last one in setup. However, it looks like it is not the case. Th idea of this “rule architecture” is to parallelise at simulation level.

root_path = 'a'
names = ['A', 'B', 'C']

rule main:
    input:
        "final.result"
    run:
        pass

rule setup:
    output:
        ABC = expand(root_path + "/{name}/prod.abc", name=names)
    run:
        pass

rule simulation:
    input:
        ABC = root_path + "/{name}/prod.abc"
    output:
        "final.result"
    run:
        pass

snakemake --dag

The last command will return:

Building DAG of jobs...
WildcardError in rule simulation in file /Users/klimt/test/Snakefile, line 19:
Wildcards in input files cannot be determined from output files:
'name'