Run IMP with testdataset

I tried to run IMP with the test dataset obtained from here.

My config file looks like this:

{
    "threads": 24,
    "memory_total_gb": 200,
    "memory_per_core_gb": 200,
    "outputdir": "/data/testdata/output/",
    "raws": {
      "Metagenomics": "/data/testdata/raw_data/metagenomics/",
      "Metatranscriptomics": "/data/testdata/raw_data/metatranscriptomics/"
    },
    "assembler": "idba",
    "vizbin": {
        "perp": 10,
        "cutoff": 1
    }
}

$ tree testdata/
testdata/
├── config.json
└── raw_data
    ├── metagenomics
    │   ├── mg.r1.fq
    │   └── mg.r2.fq
    └── metatranscriptomics
        ├── mt.r1.fq
        └── mt.r2.fq

If I try to run IMP, I get the following error:

$ impy -c /data/testdata/config.json run                                                                               
You should provide `metagenomics` and `metatranscriptomics` data.
Aborted!

Am I doing something wrong to access the config file?

(I am aware that I don't need 24 threads and 200GB of memory for this dataset, but I have problems with another dataset and I am trying to find out if the computational ressources are a problem).