maybe mis-assembly of long tandem repeats

[zhangrengang]

Here is an example:

The coverage depths of A, B, B' and C are even, so the structure is expected to be A-B-B'-C or A-B'-B-C where B and B' is heterozygous tandem repeat units (~200 kb, de=~0.001 by minimap2) and A and C is their flank sequences. But hifiasm outputs two contigs A-B-C (linear) and B' (circular) instead of A-B-B'-C. I think A-B-B'-C may be better as it is more continous and more real.

[chhylp123]

I see, thanks a lot. The UL-integration should be able to deal with this case reliably.