This repository was created to benchmark the SKYPE pipeline using the supplementary data released with the Severus manuscript. The benchmark uses the published cell-line data, structural-variant calls, and truth sets to evaluate SKYPE alongside other structural-variant callers.
The supporting dataset is available on Zenodo: Supporting data for the Severus manuscript.
skype_bench.py runs native SKYPE first for each selected cell line and saves
its successful graph-limit pair as limit_combinations.json. Every subsequent
VCF-mode case receives that file through stage 02's --limit_combinations
option. A VCF case that cannot build the graph with the native pair is marked
failed; stage 02 does not try a fallback pair.