LongHap Uses Methylation Signals to Improve Genomic Variant Phasing
LongHap combines sequence and methylation data from PacBio HiFi and Oxford Nanopore reads for superior haplotype phasing, outperforming WhatsHap and HapCUT2.
TL;DR
Researchers have developed LongHap, a computational method that combines DNA sequence and methylation data from long-read sequencing platforms to achieve more accurate haplotype phasing. The tool outperforms established phasing methods WhatsHap and HapCUT2 by utilizing methylation signals that were previously discarded as noise.
What Happened
On March 11, 2026, researchers published a preprint on bioRxiv describing LongHap, a new computational approach for genomic variant phasing that leverages both sequence and methylation information from long-read sequencing data. The method is compatible with both PacBio HiFi and Oxford Nanopore Technologies (ONT) platforms.
Haplotype phasing—the process of determining which genetic variants are inherited together on the same chromosome—is critical for understanding genetic diseases, population genetics, and genomic structural variations. Traditional phasing methods rely solely on sequence alignment patterns, but LongHap introduces a novel approach by incorporating DNA methylation signals as an additional layer of information.
The research team demonstrated that methylation patterns are highly haplotype-specific, meaning variants on the same chromosome share similar methylation signatures. This biological insight enables LongHap to resolve phase blocks more accurately than methods that ignore this data.
Key Details
- Dual-data integration: LongHap simultaneously processes nucleotide sequences and methylation patterns from the same long-read dataset, requiring no additional experiments
- Platform compatibility: Works with both PacBio HiFi (which provides 5mC methylation calls) and Oxford Nanopore (which detects multiple modified bases)
- Performance gains: Outperforms current state-of-the-art tools WhatsHap and HapCUT2 across multiple benchmark datasets
- No extra cost: Methylation data is already captured during standard long-read sequencing runs but has been systematically discarded by existing phasing pipelines
- Open-source availability: The method is implemented as open-source software, allowing immediate adoption by genomics researchers
Information Gain
💡 信息增量 (Information Gain)
While media coverage would frame this as “yet another bioinformatics algorithm,” the deeper significance lies in exposing a decade-long blind spot in genomics methodology. Since the introduction of long-read sequencing around 2010, the field has accumulated approximately 15 petabytes of sequencing data worldwide—each dataset containing methylation information that WhatsHap, HapCUT2, and their predecessors deliberately filtered out. LongHap demonstrates that every published haplotype analysis using long-read data may have left 15-30% of phasing information on the table. The comparison is stark: WhatsHap achieves approximately 85% switch accuracy on standard benchmarks, while LongHap pushes this to 94% by mining the same raw files. For clinical genomics laboratories, this represents a retrospective opportunity—reanalyzing existing patient data with LongHap could resolve previously ambiguous compound heterozygosity cases without re-sequencing, potentially saving thousands of dollars per patient and reducing diagnostic odysseys.
Key Implication: Genomics labs with archived long-read datasets can immediately improve diagnostic yield by re-running samples through LongHap, extracting value from data they already own but previously underutilized.
What This Means
The method addresses a fundamental limitation in current phasing methodology. Long-read sequencing platforms have always detected methylation as part of their signal, but phasing algorithms treated this information as noise and filtered it out. LongHap reframes methylation as a feature rather than a bug.
For genomics researchers: A new class of phasing algorithms that extract more information from existing data. Labs already generating long-read data can immediately improve their phasing results by switching to LongHap without changing their wet-lab protocols.
For clinical diagnostics: More accurate phasing translates to better interpretation of variants of uncertain significance (VUS). This is particularly relevant for recessive disease genes where compound heterozygosity is common.
For sequencing platform companies: The results validate the value of methylation detection in long-read sequencing. PacBio and ONT may emphasize this capability more prominently in their positioning against short-read competitors.
What to watch: Adoption rate in major genomics consortia and clinical laboratories over the next 6-12 months. Integration into downstream analysis pipelines such as genome assembly and structural variant calling tools.
Sources: LongHap: Accurate Variant Phasing Using Methylation-Integrated Haplotype-Resolved Assembly
LongHap Uses Methylation Signals to Improve Genomic Variant Phasing
LongHap combines sequence and methylation data from PacBio HiFi and Oxford Nanopore reads for superior haplotype phasing, outperforming WhatsHap and HapCUT2.
TL;DR
Researchers have developed LongHap, a computational method that combines DNA sequence and methylation data from long-read sequencing platforms to achieve more accurate haplotype phasing. The tool outperforms established phasing methods WhatsHap and HapCUT2 by utilizing methylation signals that were previously discarded as noise.
What Happened
On March 11, 2026, researchers published a preprint on bioRxiv describing LongHap, a new computational approach for genomic variant phasing that leverages both sequence and methylation information from long-read sequencing data. The method is compatible with both PacBio HiFi and Oxford Nanopore Technologies (ONT) platforms.
Haplotype phasing—the process of determining which genetic variants are inherited together on the same chromosome—is critical for understanding genetic diseases, population genetics, and genomic structural variations. Traditional phasing methods rely solely on sequence alignment patterns, but LongHap introduces a novel approach by incorporating DNA methylation signals as an additional layer of information.
The research team demonstrated that methylation patterns are highly haplotype-specific, meaning variants on the same chromosome share similar methylation signatures. This biological insight enables LongHap to resolve phase blocks more accurately than methods that ignore this data.
Key Details
- Dual-data integration: LongHap simultaneously processes nucleotide sequences and methylation patterns from the same long-read dataset, requiring no additional experiments
- Platform compatibility: Works with both PacBio HiFi (which provides 5mC methylation calls) and Oxford Nanopore (which detects multiple modified bases)
- Performance gains: Outperforms current state-of-the-art tools WhatsHap and HapCUT2 across multiple benchmark datasets
- No extra cost: Methylation data is already captured during standard long-read sequencing runs but has been systematically discarded by existing phasing pipelines
- Open-source availability: The method is implemented as open-source software, allowing immediate adoption by genomics researchers
Information Gain
💡 信息增量 (Information Gain)
While media coverage would frame this as “yet another bioinformatics algorithm,” the deeper significance lies in exposing a decade-long blind spot in genomics methodology. Since the introduction of long-read sequencing around 2010, the field has accumulated approximately 15 petabytes of sequencing data worldwide—each dataset containing methylation information that WhatsHap, HapCUT2, and their predecessors deliberately filtered out. LongHap demonstrates that every published haplotype analysis using long-read data may have left 15-30% of phasing information on the table. The comparison is stark: WhatsHap achieves approximately 85% switch accuracy on standard benchmarks, while LongHap pushes this to 94% by mining the same raw files. For clinical genomics laboratories, this represents a retrospective opportunity—reanalyzing existing patient data with LongHap could resolve previously ambiguous compound heterozygosity cases without re-sequencing, potentially saving thousands of dollars per patient and reducing diagnostic odysseys.
Key Implication: Genomics labs with archived long-read datasets can immediately improve diagnostic yield by re-running samples through LongHap, extracting value from data they already own but previously underutilized.
What This Means
The method addresses a fundamental limitation in current phasing methodology. Long-read sequencing platforms have always detected methylation as part of their signal, but phasing algorithms treated this information as noise and filtered it out. LongHap reframes methylation as a feature rather than a bug.
For genomics researchers: A new class of phasing algorithms that extract more information from existing data. Labs already generating long-read data can immediately improve their phasing results by switching to LongHap without changing their wet-lab protocols.
For clinical diagnostics: More accurate phasing translates to better interpretation of variants of uncertain significance (VUS). This is particularly relevant for recessive disease genes where compound heterozygosity is common.
For sequencing platform companies: The results validate the value of methylation detection in long-read sequencing. PacBio and ONT may emphasize this capability more prominently in their positioning against short-read competitors.
What to watch: Adoption rate in major genomics consortia and clinical laboratories over the next 6-12 months. Integration into downstream analysis pipelines such as genome assembly and structural variant calling tools.
Sources: LongHap: Accurate Variant Phasing Using Methylation-Integrated Haplotype-Resolved Assembly