Snakemake ancient tag with wildcards - workflow

Question

I have few SRA files which I download from NCBI website. Now I want to add them to my snakemake workflow. However, I want to retain ability to download them with prefetch if they are not available. I had following simple rule,
BASE = "/path/to/working/folder"
rule all:
input: [f"{BASE}/fastq/SRR000001.sra_1.fastq", f"{BASE}/fastq/SRR000001.sra_2.fastq"]
shell:
"echo Finished"
rule get_sra:
input: ancient("config/config.yaml")
output:"{BASE_FOLDER}/sra/{SSR_ID}.sra"
shell:
"prefetch -p {wildcards.SSR_ID} --output-file {output} "
rule get_fastq:
input: expand("{folder}/sra/{srr}.sra", folder=BASE, srr="{SRR_ID}")
output:
expand("{folder}/fastq/{srr}.sra_{i}.fastq", folder=BASE,
srr="{SRR_ID}", i=[1, 2])
shell:
"fasterq-dump {input} --outdir {BASE}/fastq"
If I use above rule, my workflow will recreate my SRA files as their timestamp will be older. However, I do not want to download full SRA file again from the server and use the already downloaded one.
For this purpose I am trying to use the ancient tag. But I am not able to use this tag with any of the wildcards.
input: ancient("{BASE_FOLDER}/sra/{SSR_ID}.sra")
Above rule gives error as
Wildcards in input files cannot be determined from output files:
Any solution to this problem? This also does not work when I use expand.

Answer 1

The problem is that not everything that you specify in curly braces is actually a wildcard. You may have 3 different use cases where you may use the curly braces:
expand functon
f-string
wildcards
In the first two cases (expand and f-string) the result is a fully specified string without any wildcards at all. If you have something like that:
rule dummy:
input: "{wildcard}.input"
output: expand("{wildcard}.output", wildcard=["1", "2"])
the result would be simply:
rule dummy:
input: "{wildcard}.input"
output: ["1.output", "2.output"]
As you can see, there are no wildcards in the output section at all, so the input cannot determine the value for it's wildcard.
The typical solution is to separate this rule into two rules:
rule all:
input: expand("{wildcard}.output", wildcard=["1", "2"])
rule do_some_work:
input: "{wildcard}.input"
output: "{wildcard}.output"
Note however that something that I called {wildcard} in the rule all: is not a wildcard per se but just an arbitrarily selected name in the local context of the expand function.