toil-vg call has become increasingly convoluted as the underlying vg commands have changed and evolved. There are too many options and possible control flows to maintain, especially when they are pretty much all obsolete. For example, when toil-vg call was written, vg call couldn't be run on a graph with much more than 10 megabases. Now it scales to a whole human genome. So this is a near-total rewrite and simplification. Some changes:
only support chunking based on connected components. (ie get rid of path-subrange and id-range chunking)
make toggling chunking on and off easier
make toggling between calling and genotyping easier
only support packed calling (as implemented in latest vg call)
do not support vg genotype
support different input and intermeidate formats (such as packed graph)
toil-vg call
has become increasingly convoluted as the underlying vg commands have changed and evolved. There are too many options and possible control flows to maintain, especially when they are pretty much all obsolete. For example, whentoil-vg call
was written,vg call
couldn't be run on a graph with much more than 10 megabases. Now it scales to a whole human genome. So this is a near-total rewrite and simplification. Some changes:vg call
)vg genotype
vg augment
toil-vg chunk
andtoil-vg augment
subcommandsShould resolve #754, #737, #519, #416, #191