odgi degree result seems inconsistent with graph topology

[adadiehl]

Greetings,
I have been working with a pangenome consisting of 51 phased genomes, two unphased genomes and GRCh38, with T2T-CHM13 as the reference, produced with Minigraph-Cactus. In trying to learn more about features in the graph, I have been exploring the VCF records for bubbles in the MC graph with various ODGI commands. Among these, I have been looking into node degree of the nodes bordering bubbles described in the VCF. However, some results seem inconsistent with my expectations based on my understanding of graph theory. For example, the following bubble with the (truncated) VCF record, bordered by nodes 4387 and 4650:

chr1 8 >4387>4650 AACCCTAACCCCTAACCCT

odgi degree -i chr1.og -g 4387
#node.id        node.degree
4387    3

odgi degree -i chr1.og -g 4650
#node.id        node.degree
4650    2

My expectation would be that both nodes should have degree >=3, since, as the boundaries of a bubble, the first node should have in-degree >=1 and out-degree >=2, and the second node should have in-degree >=2 and out-degree >=1. The first node in the example VCF record follows the expected pattern (to the best of my ability to tell, since odgi degree does not break down in-degree and out-degree), but the second node does not. Instead, a degree of two suggests only one incoming and one outgoing edge, i.e., not bordering a bubble. Can someone please explain what might be going on here? Is the bug/feature here part of ODGI, or is it likely a bug/feature of the MC graph and/or VCF?

Thank you in advance!