Most organisms are diploid, meaning they carry two sets of chromosomes—one from each parent. Yet, for decades, the "reference genomes" used in biology have been largely fictional. Traditional assembly methods typically "collapse" these two distinct sequences into a single consensus sequence. This process creates a mosaic that does not exist in nature, effectively blurring the unique genetic contributions of the mother and father into a single, artificial average.
Haplotype-resolved assembly (or "phasing") is the technical effort to stop this blurring. It aims to reconstruct the two parental sequences—the haplotypes—separately and completely.
The Consensus Illusion and the Repeat Problem
The fundamental challenge in genome assembly is that DNA sequencers do not read entire chromosomes; they read small fragments. Assemblers then try to overlap these fragments like a massive jigsaw puzzle. In a diploid organism, the maternal and paternal sequences are nearly identical, often differing by less than 1%.
As Xiongbin Kang and Alexander Schönhuth point out in their work on phasebook, traditional assemblers treat these small differences as sequencing errors or "noise." To simplify the computation, they merge the two paths into one. This "collapsed" assembly loses critical information about structural variations and heterozygosity.
The problem is compounded by genomic repeats. As Govinda M. Kamath and David Tse demonstrate with HINGE, if a repeat is longer than the sequence read, the assembler cannot tell which copy of the repeat belongs where. In a diploid context, if the assembler cannot distinguish between the maternal and paternal versions of a sequence, it produces a "chimeric" result—a sequence that jumps back and forth between the two parents.
Bridging the Gap with Long Reads
The shift toward true haplotype resolution has been driven by "third-generation" long-read sequencing (such as PacBio and Oxford Nanopore). Because these reads are tens of thousands of base pairs long, they can span multiple genetic variants. If a single read contains two different variants, we know for a fact those variants sit on the same physical chromosome.
Modern pipelines like PECAT, developed by Fan Nie and Chuan‐Le Xiao, use these long, "noisy" reads to perform haplotype-aware error correction. Instead of averaging out differences, the algorithm identifies clusters of reads that share the same variants and groups them into separate "bins" for each haplotype before the final assembly begins. This ensures that the maternal and paternal sequences are never mixed in the first place.
Trio-Based vs. Trio-Free Phasing
How do we know which sequence belongs to which parent? The most robust method is "trio-based" assembly, which uses sequencing data from the parents to sort the offspring's reads. Erich D. Jarvis and the Human Pangenome Reference Consortium have used this to produce near-complete diploid human genomes with only a handful of gaps per chromosome.
However, parental data is often unavailable, especially in wild species. Researchers like Liqun Han have pioneered trio-free assembly in organisms like the wild walnut. This approach relies on "chromatin conformation" data (Hi-C). Because chromosomes are physically separated in the cell nucleus, Hi-C captures interactions between DNA segments that are close to each other in 3D space. Since maternal and paternal chromosomes rarely touch, these spatial "contacts" act as a scaffold to link distant variants together on the same haplotype. Chuan‐Le Xiao further refined this with Dip3D, which uses nanopore "Pore-C" data to reconstruct high-order 3D interactions, significantly increasing the accuracy of the phasing.
Why Resolution Matters
The move from collapsed to phased genomes is not just a technical refinement; it reveals biological "dark matter" that was previously invisible.
- The Immune System: The Major Histocompatibility Complex (MHC) is the most diverse part of the mammalian genome. Collapsed assemblies fail here because the maternal and paternal versions are too different to merge. Tingting Li used haplotype-aware assembly to finally resolve the bovine immune system, uncovering the structural secrets of cattle antibodies.
- Evolutionary History: By resolving the genomes of six ape species from telomere-to-telomere, DongAhn Yoo and Arang Rhie were able to see gene-rich duplications and centromeres that define primate evolution.
- Allele-Specific Expression: In many diseases, including cancer, only one of the two inherited copies of a gene might be active. Without a phased genome, it is impossible to tell which copy is being expressed. Claudia Calabrese and colleagues have shown that allele-specific regulatory landscapes are major drivers of tumor dysregulation.
By treating the genome as a diploid reality rather than a haploid abstraction, researchers are moving toward a more precise understanding of how inheritance, structure, and 3D shape combine to drive biology.