QuaC fails when only one sample in input ped file

Rules somalier_relate, somalier_ancestry and mosdepth_plot fail when input ped file has only sample listed. This is due to filename pattern globbing used in these rules to restrict input samples to only those specified in input.

Potential solutions dicussed with @bwilk777 :

1. Change logic in workflow to use globbing only when there are >1 samples
2. Do not depend on globbing at all. Instead just list out each input file. For example: some_command file1 file2 file3 .....

There is no limit (other than available memory) to the number of files that may be expanded by a bash glob. However when those files are passed as arguments to a command that is executed (as opposed to a shell builtin or function), then you may run into a limit of the execve() system call on some systems.

Source

Based on this, globbing may not necessarily be a great idea.

We will therefore go with solution-2, which is not to use globbing at all. Earlier I didn't want to use this setup as snakemake would run into error with long shell commands. However in versions since 6.0.0, snakemake appears to have introduced a fix for this. As we use snakemake 6.0.5, we probably will never have to worry about this.

mentioned in issue #35 (closed)

Removed globbing and then tested with CF-T** project with 82 samples. Somalier relate command was ~9900 characters long, and it ran successfully. Note that this appears to be within the allowed shell command length though.

mentioned in commit 9b626476

mentioned in commit 1e884e0e

mentioned in merge request !3 (merged)

mentioned in commit fdbbd797

closed with merge request !3 (merged)