minigraph-cactus inherits its large-scale topology from minigraph, with the reference genome kept acyclic and extra copies of things being added as novel insertion sequences. This is great for keeping a simple coordinate system, especially with regards to projecting to the reference. But it can confuse mapping a bit.
This PR adds some prototype options for more aggressively collapsing similar sequence, adding cycles as necessary. They are applied to cactus-pangenome or cactus-graphmap:
--collapse reference : Incorporate self-alignments on the reference genome as generated by minimap2 -c -PD -k19 -w19 -m200 (options can be changed via minimapCollapseOptions in the config). Note that only alignments within (and not between) contigs are used
--collapse all : As above, but run self-alignment on all input genomes
--collapseRefPAF <paffile> : Use self-alignments from given PAF file (only on reference genome).
If running on cactus-graphmap, then the --collapseRef option must be given to cactus-graphmap-join.
Update: collapsing simplified to a single boolean --collapse flag. reference / non-reference collapsing can still be experimented with, but only via the config XML.
minigraph-cactus inherits its large-scale topology from minigraph, with the reference genome kept acyclic and extra copies of things being added as novel insertion sequences. This is great for keeping a simple coordinate system, especially with regards to projecting to the reference. But it can confuse mapping a bit.
This PR adds some prototype options for more aggressively collapsing similar sequence, adding cycles as necessary. They are applied to
cactus-pangenomeorcactus-graphmap:--collapse reference: Incorporate self-alignments on the reference genome as generated byminimap2 -c -PD -k19 -w19 -m200(options can be changed viaminimapCollapseOptionsin the config). Note that only alignments within (and not between) contigs are used--collapse all: As above, but run self-alignment on all input genomes--collapseRefPAF <paffile>: Use self-alignments from given PAF file (only on reference genome).If running on
cactus-graphmap, then the--collapseRefoption must be given tocactus-graphmap-join.Update: collapsing simplified to a single boolean
--collapseflag. reference / non-reference collapsing can still be experimented with, but only via the config XML.