[Draft] New command: "hybracter automatic" - hybrid and long samples in one go!

[richardstoeckl]

Hi George,

In my workflow, I sometimes have samples that have short read data available for them and sometimes there is only long read data available.
So every time I want to assemble them, I need to execute hybracter twice, instead of just letting it run over night in one go.

So I was wondering, why you separate the "hybrid" and "long-only" mode so strictly? Is this a deliberate design choice?

In this PR I wanted to present a mockup (!!) of how an "automatic" mode could work, which runs either the hybrid or the long-only pipeline, depending on the availability of short read paths in the sample.csv file (as seen in hybracter/test_data/test_hybrid_automatic_auto.csv), on a per-sample basis.

Basically, the idea is to add some logic to hybracter/workflow/rules/preflight/samples.smk, so that the SAMPLES can be split into HYBRID_SAMPLES and LONG_SAMPLES, which can then be used in wildcard_constraints and in expand() functions to choose which samples get processed by either pipeline.

Obviously, this disables some sanity checks regarding the required existence of paths/files, but maybe one could communicate this mode to be best suited for experienced people?

For this PR, I just did a very crude mockup, as I did not want to waste too much time if this Idea is not something you want to pursue further.
Please excuse the horrible python code in samplesFromCsvAutomatic(), parseSamples(), and filter_samples_by_workflow_type() and the copy-and-pasted-together rules and snaketool code, I tried my best to understand the inner workings.
The mockup works only with hybracter automatic -i hybracter/test_data/test_hybrid_automatic_auto.csv --no_medaka --databases /path/to/databases/ --auto, again, to not waste too much time. It "works" in the sense that it runs through on my machine without errors, but I have not checked for mistakes or on other machines.

If you want to pursue this idea further, let me know, and I can polish everything up a bit more.

Best wishes,
Richard

PS: "hybracter automatic" is probably not the best name, especially with regards to confusion about the --auto parameter, but I couldn't think of something better for now.