- Haplotype-aware and diploid genome assembly are approaches used to reconstruct genomes while accounting for the genetic differences between homologous chromosomes. In diploid organisms, such as humans, most mammals, and many plants, each cell typically contains two copies of each chromosome: one inherited from the maternal parent and the other from the paternal parent. Although these chromosome copies contain largely corresponding genes and genomic regions, they may differ at many nucleotide positions and may also contain different insertions, deletions, duplications, and structural variations. These differences make diploid genome assembly more complex than assembling a genome from a haploid or genetically uniform sequence.
- Traditional genome assembly workflows may combine sequencing reads from both chromosome copies into a single consensus representation. This can produce a useful reference-like genome, but it may conceal differences between the two homologous chromosomes. Haplotype-aware assembly attempts to preserve these differences by identifying which sequencing reads belong to each chromosome copy and reconstructing the corresponding sequences. The resulting assemblies may represent two separate haplotypes, partially phased chromosome sequences, or a primary assembly accompanied by alternate haplotigs.
- A haplotype is a combination of genetic variants located on the same physical chromosome or DNA molecule. In a diploid organism, the two homologous chromosomes carry two haplotypes for most autosomal regions. For example, an individual may carry one allele of a gene on the maternal chromosome and a different allele on the paternal chromosome. If several heterozygous variants occur close together, knowing which variants are present on the same chromosome is important for understanding their biological relationships. This information is known as phase. Two variants are in cis when they occur on the same chromosome and in trans when they occur on opposite homologous chromosomes.
- Diploid genome assembly aims to reconstruct the genomic sequences of both chromosome copies rather than treating them as one undifferentiated sequence. A fully phased diploid assembly attempts to represent both haplotypes separately across the genome. In practice, complete phasing may not be possible in every region because of insufficient sequencing coverage, low heterozygosity, long identical-by-descent regions, repetitive DNA, or complex structural variation. Assemblies may therefore contain fully phased regions, partially phased regions, and unresolved sequences.
- Haplotype-aware assembly is particularly important in human and animal genomics, plant breeding, population genetics, evolutionary research, and the study of inherited diseases. It can help identify allele-specific differences, characterize structural variants, investigate gene regulation, study recombination, and understand how genetic variation is organized across chromosomes. In agricultural genomics, haplotype information can support the analysis of economically important traits, genetic diversity, inherited disorders, and breeding values. In clinical genomics, it can help interpret compound heterozygous variants, identify disease-associated haplotypes, and investigate allele-specific genetic effects.
- The main challenge in diploid assembly arises from heterozygosity. Heterozygous positions contain different nucleotides on the two homologous chromosomes. For example, one chromosome may carry an adenine at a particular position, while the other carries a guanine. If sequencing reads from both chromosomes are combined without preserving their origin, the assembler may interpret the differences as alternative sequence paths or may collapse them into a single consensus representation. This can lead to loss of haplotype-specific information.
- In an assembly graph, heterozygous differences may appear as bubbles. A bubble is a region in which two alternative paths connect the same or corresponding graph positions. These paths may represent alternative alleles on homologous chromosomes. However, not every bubble represents a simple heterozygous variant. Some may arise from sequencing errors, repetitive sequences, paralogous regions, or structural differences. Assemblers use read support, sequence similarity, coverage, and other information to distinguish likely biological variation from technical artifacts.
- Heterozygosity can also produce an assembly problem known as haplotype collapse. This occurs when highly similar sequences from the two homologous chromosomes are combined into one assembled sequence. The resulting assembly may represent a consensus or one predominant sequence, but it no longer preserves both chromosome copies independently. Collapsed regions may have higher read depth because reads from both homologues map to the same assembled sequence. They can also conceal allele-specific insertions, deletions, or structural differences.
- The opposite problem is haplotype duplication, sometimes called redundant haplotig retention. In this situation, both homologous chromosome sequences are represented as separate contigs even when the intended assembly is a single primary representation. This can inflate the total assembly length and create apparent duplicated genes or genomic regions. Whether retaining both copies is an error depends on the assembly objective. A haplotype-resolved diploid assembly should preserve both chromosome copies, whereas a primary or haploid-representation assembly generally aims to avoid unnecessary duplication.
- Long-read sequencing has substantially improved haplotype-aware assembly because long reads can span multiple heterozygous variants. When a read contains several variants, it provides evidence that those variants occur on the same physical DNA molecule. This helps determine phase across a genomic region. PacBio HiFi reads are particularly useful because they combine relatively long read lengths with high consensus accuracy. Oxford Nanopore reads can also provide long-range phasing information, especially when reads span multiple heterozygous sites or structural variants. Their suitability depends on read accuracy, coverage, and the basecalling and assembly methods used.
- Short-read sequencing can also contribute to phasing, particularly when paired-end reads or linked information connect nearby heterozygous variants. However, ordinary short reads usually provide limited long-range phasing because they cover relatively small genomic intervals. Statistical phasing methods can use population reference panels and patterns of linkage disequilibrium to infer likely haplotypes, but these are predictions rather than direct observations of the physical chromosome sequence. Long-read sequencing, linked-read technologies, chromosome conformation data, and parental sequencing can provide additional evidence for longer-range phasing.
- Trio-binning is an important strategy for haplotype-resolved assembly when sequencing data are available from an offspring and both biological parents. The method uses inherited genetic differences to assign offspring reads to maternal and paternal groups. Reads carrying variants inherited from one parent are assigned to the corresponding parental haplotype, while reads carrying variants inherited from the other parent are assigned to the second haplotype. The two read groups can then be assembled separately, reducing the complexity of reconstructing the diploid genome.
- Trio-binning is especially useful for highly heterozygous genomes and can produce separate maternal and paternal assemblies. Its effectiveness depends on the availability and quality of parental sequencing data, the number of informative variants, and the ability to assign reads correctly. Regions with little parental genetic differentiation may be difficult to phase, while sequencing errors or incorrect parental data can cause misassignment. Trio-binning also requires careful handling of sex chromosomes, mitochondrial DNA, and regions with unusual inheritance patterns.
- Another approach is read-based phasing, in which sequencing reads are used to connect heterozygous variants into phase blocks. If two variants occur within the same read, the read provides direct evidence of their chromosome association. When reads overlap or connect through intermediate variants, the phasing information can extend beyond the length of individual reads. A phase block is a genomic region in which variants have been assigned to haplotypes with supporting evidence. Separate phase blocks may remain unconnected when there is insufficient information to establish their relative orientation.
- Hi-C sequencing can provide additional long-range information for phasing and chromosome organization. Hi-C captures DNA regions that are physically close to one another in the nucleus. These interactions can support the assignment of contigs to chromosomes and, in suitable workflows, help distinguish homologous chromosome copies. However, Hi-C data do not directly sequence the entire chromosome, and their interpretation requires specialized computational methods. Hi-C evidence is often combined with long reads and other information to improve chromosome-scale phasing and scaffolding.
- The assembly of diploid genomes commonly involves several computational stages. First, sequencing reads are generated using suitable short-read, long-read, or combined technologies. The reads are then assessed for quality, filtered where necessary, and examined for coverage and contamination. The assembler identifies overlaps or graph relationships among the reads and constructs an initial assembly representation. Heterozygous variants and alternative sequence paths are identified, and phasing methods assign supported reads or contigs to haplotypes. The resulting sequences are assembled, polished, and evaluated for completeness, accuracy, and correct separation of the two chromosome copies.
- Several software tools are used for haplotype-aware and diploid assembly. Hifiasm is widely used for assembling PacBio HiFi reads and supports haplotype-aware assembly, including trio-binning workflows. HiCanu is designed for high-quality long-read assembly and can be used in suitable diploid assembly workflows. FALCON-Unzip was developed to generate primary contigs and associated haplotigs from long-read data, preserving some haplotype-specific sequence information. Verkko is designed for complex genome assembly and can use long-read and other supporting data to improve the reconstruction of highly contiguous and phased genomes. Other tools, including Flye and Canu, are useful for long-read assembly, although their haplotype handling and output representations differ.
- Hifiasm is particularly important in modern diploid genome assembly because it uses PacBio HiFi reads to construct genome assembly graphs and identify haplotype-specific sequence paths. It can generate primary and alternate assembly representations, and it supports trio-binning when parental sequencing data are available. A basic assembly command is:
hifiasm \
-o diploid_assembly \
-t 16 \
reads.hifi.fastq.gz
- This command performs a standard Hifiasm assembly using HiFi reads. The output includes assembly graph files and sequence representations that should be interpreted according to the software documentation. A primary assembly is not necessarily a completely phased haplotype, and alternate sequences may represent additional haplotype-specific regions.
- When parental sequencing data are available, Hifiasm can perform trio-binning using parental k-mer information. The exact workflow requires the appropriate parental read sets and compatible software options. A simplified command structure is:
hifiasm \
-o trio_assembly \
-t 16 \
-1 parent1.reads.fastq.gz \
-2 parent2.reads.fastq.gz \
offspring.hifi.fastq.gz
- Here, parent1 and parent2 represent the two biological parents, and the offspring reads are assembled using parental information. The example illustrates the general input structure; researchers should verify the exact options and supported input formats in the installed Hifiasm version.
- FALCON-Unzip is another approach designed to separate haplotype-specific sequences during assembly. It can produce primary contigs and associated haplotigs, which represent alternative sequences corresponding to heterozygous genomic regions. The primary contigs provide a principal representation of the genome, while the associated haplotigs preserve additional haplotype information. The interpretation of these outputs depends on the assembly objective and the degree of phasing achieved.
- Verkko can be used for highly contiguous genome assembly and supports workflows that integrate different types of long-range sequencing information. It is particularly relevant for complex genomes where resolving repeats, haplotypes, and chromosome structures is important. The exact input requirements depend on the available read types and supporting data, so the workflow should be configured according to the software documentation and sequencing design.
- Assembly polishing is an important step in diploid genome reconstruction. Polishing aims to correct errors in the assembled nucleotide sequences by using supporting sequencing reads. However, haplotype-aware polishing requires special care. Reads from the two homologous chromosomes should not be indiscriminately mixed when they contain genuine allele-specific differences. If reads from both haplotypes are used to correct one haplotype-specific sequence, true heterozygous variants may be incorrectly converted into consensus bases. Suitable workflows should therefore preserve haplotype assignments or use polishing methods designed for the relevant assembly representation.
- Haplotype-aware assembly quality assessment must evaluate both sequence quality and phasing quality. General assembly statistics include total sequence length, number of contigs, largest contig, N50, L50, GC content, and read coverage. These measures describe assembly properties but do not prove that maternal and paternal sequences have been correctly separated. A high N50, for example, indicates contiguity rather than accurate phasing.
- BUSCO can be used to evaluate the completeness of conserved gene sets in the assembly. It can also help identify possible duplication patterns, although duplicated BUSCO genes may be biologically expected in a diploid assembly that retains both homologous chromosome copies. Interpretation must therefore consider whether the objective is a haploid representation or a fully diploid, haplotype-resolved assembly.
- Merqury is a tool used for reference-free genome assembly evaluation based on k-mer comparisons between sequencing reads and the assembly. It can estimate consensus quality, completeness, and k-mer representation. In suitable workflows, it can also provide information relevant to haplotype evaluation. Other tools, including QUAST, can assess assembly statistics and compare sequences with an appropriate reference genome. Reference-based comparisons must be interpreted carefully because genuine differences between the sample and reference can appear as discrepancies.
- Phasing quality can be evaluated using metrics such as switch error rate, phase-block length, and haplotype completeness. A switch error occurs when the inferred phase changes incorrectly between adjacent heterozygous variants. Phase-block length describes the genomic distance over which variants have been assigned to a consistent haplotype group. Longer phase blocks are generally useful, but their quality depends on the accuracy of the underlying variant assignments. Parental information, read-based evidence, and independent sequencing technologies can help validate phasing.
- Read mapping is also important for detecting possible haplotype collapse, duplication, or incorrect sequence assignment. If reads from both homologues map to a single assembled region, the coverage may be higher than expected for a separately assembled haplotype. If homologous sequences are represented separately, reads may show more distinct alignment patterns. However, coverage alone cannot establish whether a region is correctly phased because sequencing bias, copy-number variation, repetitive DNA, and mapping ambiguity can produce similar patterns.
- Haplotype-aware assembly has important applications in human genetics. Many inherited diseases involve variants whose interpretation depends on whether they occur on the same or opposite chromosomes. For example, some recessive disorders require pathogenic variants to be present on both copies of a gene. Determining whether two variants are in cis or trans can therefore affect clinical interpretation. Haplotype-resolved assemblies can also help investigate structural variants, repeat expansions, allele-specific expression, and regions that are difficult to characterize using conventional reference-based methods.
- In animal and plant breeding, haplotype information can support the identification and tracking of beneficial alleles across generations. A haplotype may contain multiple linked variants associated with economically important traits, disease resistance, fertility, growth, or environmental adaptation. Haplotype-aware genomic analysis can help researchers understand inheritance patterns and linkage relationships, improve genetic characterization, and support genomic selection. However, an association between a haplotype and a trait does not automatically demonstrate causation; validation in appropriate populations and environments remains necessary.
- In evolutionary biology, diploid assembly can help reconstruct the two inherited chromosome copies and examine how they differ. This is useful for studying recombination, population history, hybridization, gene duplication, and structural variation. Haplotype-resolved sequences can also help distinguish allelic differences from paralogous gene copies and improve comparative analyses between closely related species.
- Despite its advantages, haplotype-aware and diploid assembly has several limitations. High heterozygosity can create complex assembly graphs, while low heterozygosity may provide insufficient information to distinguish homologous chromosomes. Long repetitive regions may remain unresolved even with long reads, especially when repeats exceed the available read lengths or have highly similar copies. Uneven sequencing coverage, contamination, sequencing errors, and structural variation can also complicate phasing. In polyploid organisms, the problem becomes more complex because multiple homologous or related chromosome copies must be distinguished.
- The interpretation of assembly outputs also requires care. A primary assembly accompanied by alternate haplotigs is not necessarily equivalent to two complete, independently phased haplotype assemblies. Some primary representations contain regions where the two haplotypes have been collapsed, while alternate contigs may cover only parts of the genome. Researchers should examine the assembler’s documentation and evaluate the output structure before describing an assembly as fully phased or haplotype-resolved.
- Reproducibility is essential in diploid genome assembly. Researchers should preserve raw sequencing reads, parental data when used, assembly graphs, intermediate files, and final sequence representations. They should document sequencing platforms, library preparation methods, coverage, software versions, command-line parameters, phasing strategies, polishing procedures, and quality-control results. Checksums can be used to verify file integrity, while workflow managers such as Snakemake and Nextflow can help organize the analysis into reproducible pipelines. Conda environments and container technologies such as Docker and Apptainer can help preserve software dependencies.
- Haplotype-aware and diploid genome assembly have expanded the ability of bioinformatics researchers to reconstruct genomes without losing important information about homologous chromosome differences. By preserving maternal and paternal sequence information, these methods provide a more detailed representation of genetic variation, inheritance, and genome organization. Tools such as Hifiasm, FALCON-Unzip, HiCanu, and Verkko support different approaches to haplotype-aware reconstruction. Their successful application depends on suitable sequencing data, adequate coverage, reliable phasing information, and careful validation. These approaches are increasingly important for producing complete genome references, investigating inherited variation, and understanding the genetic architecture of complex organisms.