broadinstitute / catch / 386

94%

master: 94%

push

travis-ci-com

Add design_large.py, which sets options suited to large, diverse input

This commit adds design_large.py, which wraps design.py and sets options
and parameters suited for large, highly diverse input. The default
options (notably, `--filter-with-lsh-minhash 0.6` and
`--cluster-and-design-separately 0.15`) lower runtime and memory usage.
The downside is usually a small increase in the number of output probes.
It additionally sets parameters (notably, `--mismatches 5` and
`--cover-extension 50`) to reasonable values, rather than the
restrictive defaults (0 for each) used by design.py.

Since design_large.py simply wraps design.py, this commit also includes
minor restructuring in design.py to support that. Arguments are now
initialized based on an `args_type` that is either `basic`
(corresponding to the usual design.py) or `large` (used by
design_large.py).

The commit also adds several warnings to assist users who might run
design.py or design_large.py in an atypical manner, such as recommending
options when provided with large input.

1817 of 2068 branches covered (87.86%)

5693 of 6037 relevant lines covered (94.3%)

1.89 hits per line