TL;DR: mainly need an answer to #4 — everything else is context for why we're asking.

We're working with the officially published HPRC v2.1 minigraph-Cactus CHM13 per-chromosome release (hprc-v2.1-mc-chm13.chroms/) and need to cross-reference node IDs from GAM alignments against the full graph with node IDs in the per-chromosome .og files. We checked two files and found both start renumbering from 1:

chr1.full.og: min_node_id=1, max_node_id=19,612,454

chr10.full.og: min_node_id=1, max_node_id=9,053,079

(confirmed via odgi view -i <file> -g, streamed through awk tracking min/max node ID from the GFA S-lines)

1. Are the officially published per-chromosome .og/GFA outputs expected to share the full graph's node ID space, or is per-chromosome renumbering starting at 1 the intended/expected behavior?
2. If node IDs aren't shared: were these per-chromosome files extracted from one whole-genome graph (in which case a global -> local offset/translation should exist somewhere), or were they built independently per chromosome and joined afterward for the whole-genome release (in which case is there any global ID space at all short of building the full graph ourselves)?
3. Is there a flag or build mode we're missing that would preserve/produce global node IDs in the per-chromosome outputs?
4. Is there a manifest, .dist index, or ID-offset table meant to accompany the split files that may not have been included in v2.1? If one doesn't exist publicly, would it be possible to generate and share one?