# How to Phase a Genome with Trio Data: A Step-by-Step Guide to Using Long Reads from Parents and Offspring Trio phasing uses sequence data from both parents and one offspring to assign inherited variants to the maternal or paternal chromosome copy. Long reads from nanopore or PacBio platforms extend phase blocks across hundreds of kilobases, which enables researchers to resolve haplotypes for downstream analysis of compound heterozygosity, allele-specific expression, and structural variant interpretation. This guide provides a practical workflow for researchers who have already generated or plan to generate long-read sequencing data for a family trio. The protocol covers raw data organization, read alignment, variant calling, phasing with WhatsHap, quality assessment, and common troubleshooting steps. The intended reader is a graduate student, laboratory researcher, or bioinformatics analyst who has basic command-line experience and needs a reproducible path from raw FASTQ files to phased variant calls. ## Scope and Prerequisites This workflow assumes you have access to a Linux-based computing environment with at least 16 CPU cores and 64 GB of RAM for small genomes, with proportionally larger resources for mammalian genomes. You will need the following software installed: a long-read aligner such as minimap2, a variant caller appropriate for your platform, WhatsHap for phasing, and samtools for BAM file manipulation. The [Galaxy Training Network](https://training.galaxyproject.org/) provides accessible tutorials for each of these steps if you prefer a graphical interface over command-line execution. For researchers who need structured learning paths across multiple bioinformatics topics, the [EMBL-EBI Training portal](https://www.ebi.ac.uk/training) offers curated courses on sequence analysis and data resources. The input data requirements are straightforward. You need three FASTQ files, one for each family member, generated from the same sequencing platform. The offspring sample is the one you will phase. The parental samples provide the reference genotypes that allow WhatsHap to determine which allele in the offspring came from which parent. For best results, all three samples should be sequenced at comparable depth. The [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/) provide access to public reference genomes and annotation files that you will need for alignment and variant annotation. ## At a Glance | Workflow Stage | Primary Tool | Key Input | Main Output | Critical Quality Check | |---|---|---|---|---| | Read alignment | minimap2 | FASTQ files for trio members | BAM files with read groups | Mapping rate above 90 percent for each sample | | Variant calling | PEPPER-Margin-DeepVariant or Clair3 | BAM files for trio members | VCF files with genotype calls | Variant quality scores and transition-transversion ratio | | Phasing | WhatsHap | Offspring VCF plus parental VCFs | Phased VCF with phase blocks | Phase block N50 and switch error rate | | Validation | WhatsHap stats or custom scripts | Phased VCF and BAM files | Quality metrics report | Mendelian consistency across the trio | ## Understanding Trio Phasing Principles ### Why Parental Data Improves Phasing Accuracy Phasing is the process of determining which alleles on homologous chromosomes were inherited together from a single parent. Without parental information, phasing relies entirely on the observation that reads containing multiple heterozygous variants must originate from the same chromosome copy. This read-based phasing works well when reads are long enough to span multiple variants, but it produces fragmented phase blocks in regions of low heterozygosity. Trio data solves this problem by providing external evidence. If the mother is homozygous for the reference allele at a position and the father is homozygous for the alternate allele, then the offspring must have inherited the alternate allele from the father. WhatsHap uses these Mendelian constraints to assign alleles to parental haplotypes with high confidence. The advantage of long reads in this context is substantial. A study describing the PEPPER-Margin-DeepVariant pipeline demonstrated that nanopore reads can produce phase blocks that contiguously span between 85 and 92 percent of annotated genes across six samples tested. This level of contiguity is difficult to achieve with short reads because the physical distance between heterozygous variants often exceeds read length. Long reads bridge these gaps and allow WhatsHap to connect variants that would otherwise remain in separate phase blocks. ### The Role of Recombination and Inheritance Patterns Each offspring receives one chromosome copy from each parent, but the parental chromosomes themselves are mosaics of grandparental chromosomes due to recombination during meiosis. This means that the phase blocks you reconstruct in the offspring will not necessarily match the complete parental haplotypes. Instead, you will observe segments where the offspring inherited a contiguous block from one grandparental chromosome, followed by a switch to the other grandparental chromosome. These recombination breakpoints appear as positions where the parental origin of alleles changes. WhatsHap does not explicitly model recombination, but the phase blocks it produces will naturally reflect these breakpoints because the Mendelian constraints change at each recombination event. For researchers studying genetic disease, this distinction matters. A compound heterozygote carries two different pathogenic variants in the same gene, one on each chromosome copy. Trio phasing can determine whether these variants are in trans, meaning on opposite chromosomes, or in cis, meaning on the same chromosome. This information has direct clinical relevance because compound heterozygosity in trans is often required for recessive disease expression. The [parent-of-origin detection study using long-read methylation sequencing and Strand-seq](https://doi.org/10.1016/j.xgen.2022.100233) showed that determining whether an inherited disease allele originated from the mother or the father can improve the diagnosis and management of many genetic diseases. While that study used a different technical approach, the principle that parental origin information has diagnostic value applies directly to trio phasing workflows. ## Data Preparation and Quality Control ### Organizing Raw Sequencing Data Before you begin any analysis, establish a clear directory structure for your project. A typical layout includes separate directories for raw data, alignment files, variant calls, and phasing output. Use consistent file naming conventions that include the sample identifier, sequencing platform, and data type. For example, a file name like `NA12878_nanopore_R10.fastq.gz` immediately communicates the sample, platform, and pore version. This practice becomes essential when you work with multiple samples or revisit analyses months later. Record the sequencing depth and read length distribution for each sample before proceeding. The [The Carpentries Lessons](https://carpentries.org/lessons) provide foundational training in shell scripting and data organization that will help you automate these quality checks. Create a simple text file that documents the number of reads, total bases, N50 read length, and estimated coverage for each family member. This file serves as your project metadata and helps you diagnose problems later if alignment or variant calling produces unexpected results. ### Read Quality Assessment Run quality assessment on each FASTQ file using a tool such as NanoPlot for nanopore data or LongQC for PacBio data. These tools generate summary statistics and visualizations that reveal problems such as adapter contamination, low-quality tail regions, or unusual read length distributions. The [Galaxy Training Network](https://training.galaxyproject.org/) offers tutorials on quality control that walk through the interpretation of these metrics. Key metrics to record include the median read quality score, the N50 read length, and the proportion of reads that pass your quality threshold. For nanopore data, a median quality score above Q10 is generally acceptable for variant calling, though newer chemistry and basecalling models routinely produce Q20 or better. For PacBio HiFi data, quality scores are typically much higher, often exceeding Q30. If you observe a large proportion of short reads below 1 kilobase, this may indicate library preparation problems or excessive fragmentation during sequencing. Document these observations in your project metadata because they will help you interpret downstream results. ### Reference Genome Selection Choose the reference genome assembly that matches your organism and, for human data, the version that is current for your research community. The [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/) provide access to reference assemblies, annotation files, and the associated metadata that describe assembly quality and coordinate systems. Download the reference genome FASTA file and its associated index files. Create a dictionary file for the reference using Picard tools or samtools, as this is required by some variant callers. Record the exact reference version and the date you downloaded it in your project metadata. Reference genomes are updated periodically, and using different versions across samples or projects will cause coordinate mismatches that complicate comparative analysis. If you need to change reference versions mid-project, plan to re-align all samples instead of attempting to lift over coordinates, because liftover introduces errors and loses information. ## Read Alignment ### Aligning Long Reads with Minimap2 Minimap2 is the standard aligner for long reads and supports both nanopore and PacBio data. The alignment command differs by platform. For nanopore data, use the preset `map-ont`. For PacBio HiFi data, use the preset `map-hifi`. These presets set the appropriate scoring parameters and minimize the alignment error rate for each data type. The basic alignment command for each sample follows this pattern: ``` minimap2 -ax map-ont -t 16 reference.fasta sample.fastq.gz | samtools sort -o sample.sorted.bam ``` Add the `-R` option to assign a read group label to each sample. Read groups are essential for downstream analysis because they allow variant callers to distinguish samples in a multi-sample BAM file and enable the joint analysis of trio data. A typical read group string includes the sample ID, library ID, platform, and a unique identifier. For example: ``` -R '@RG\tID:sample1\tSM:sample1\tPL:ONT' ``` The `SM` tag must match the sample name you will use in the variant calling step. Inconsistent sample naming between the BAM file and the VCF file is a common source of errors in trio analysis. ### Post-Alignment Processing After alignment, sort the BAM file by coordinate position and index it with samtools. Then run alignment statistics to assess mapping quality. The key metric is the proportion of reads that map to the reference genome. For high-quality long-read data, you expect more than 90 percent of reads to map. Lower mapping rates may indicate contamination, adapter problems, or a mismatch between the sample and reference genome. Use `samtools stats` and `samtools flagstat` to generate alignment summaries. Record the following metrics for each sample: total reads, mapped reads, mapping rate, median insert size, and coverage. The [Bioconductor project](https://bioconductor.org/) provides R packages for visualizing these alignment statistics if you prefer to work in an R environment. The [Galaxy Training Network](https://training.galaxyproject.org/) also offers alignment tutorials that include quality assessment steps. ### Multi-Sample BAM Construction For trio phasing, you have two options. You can align each sample separately and run variant calling on each BAM independently, then combine the VCF files during phasing. Alternatively, you can merge the three sample BAM files into a single multi-sample BAM and run joint variant calling. The joint approach is generally preferred because it allows the variant caller to use information across samples to improve genotype calls at low-depth positions. To create a multi-sample BAM, align each sample separately with the appropriate read group, then use `samtools merge` to combine the three sorted BAM files. Ensure that each sample has a distinct read group ID and that the sample names in the read groups match the names you will use in the VCF. After merging, index the combined BAM file and verify that all three samples are present using `samtools view` with the sample name filter. ## Variant Calling ### Platform-Specific Variant Calling Strategies The choice of variant caller depends on your sequencing platform. For nanopore data, the PEPPER-Margin-DeepVariant pipeline has demonstrated state-of-the-art performance. The [study describing this pipeline](https://doi.org/10.1038/s41592-021-01299-w) showed that it outperforms short-read-based single-nucleotide variant identification at the whole-genome scale and produces high-quality variants in segmental duplications and low-mappability regions where short-read genotyping fails. The pipeline consists of three stages: PEPPER for initial variant detection, Margin for haplotyping, and DeepVariant for final genotype refinement. For PacBio HiFi data, the same study showed that the PEPPER-Margin-DeepVariant pipeline provides an efficient solution with superior performance over the standard WhatsHap-DeepVariant approach. Alternatively, you can use DeepVariant directly with the appropriate model for HiFi data. The [nf-core documentation](https://nf-co.re/docs) describes community-standard pipelines that implement these variant calling approaches in a reproducible manner, which is useful if you prefer to use a pre-built workflow instead of constructing your own. ### Joint Genotyping for Trio Analysis When you run variant calling on the multi-sample BAM, the variant caller produces a single VCF file with genotype calls for all three samples. This joint calling approach has several advantages for trio analysis. First, it allows the caller to use the parental genotypes to inform the offspring genotype at low-depth positions. Second, it produces a single VCF that is already coordinated across samples, which simplifies the phasing step. Third, it enables the calculation of Mendelian consistency statistics that help identify genotyping errors. After variant calling, run a quality filter to remove low-confidence calls. The specific filters depend on the variant caller and platform, but common filters include minimum genotype quality, minimum depth, and minimum variant quality score. Record the number of variants before and after filtering in your project metadata. A sudden drop in variant count after filtering may indicate that your quality thresholds are too stringent or that the sequencing depth is inadequate. ### Structural Variant Considerations Long reads are particularly valuable for detecting structural variants, which are often missed by short-read sequencing. The [PEPPER-Margin-DeepVariant study](https://doi.org/10.1038/s41592-021-01299-w) demonstrated that nanopore-based methods can produce high-quality single-nucleotide variants in segmental duplications and low-mappability regions where short-read genotyping fails. These regions are also enriched for structural variants, so the same long-read data can be used for both small variant calling and structural variant detection. For structural variant calling, tools such as Sniffles2 or cuteSV are commonly used with long-read data. These tools produce a separate VCF file for structural variants. Trio phasing of structural variants is more complex than phasing of single-nucleotide variants because structural variants often span large genomic regions and may not be represented as simple alleles. However, the phase information from small variants can be used to assign structural variants to haplotypes if they overlap with phased small variants. This approach is an active area of research, and the [haplotype-aware segmentation study](https://doi.org/10.1186/s13059-026-03971-w) demonstrates the broader utility of haplotype information for genomic analysis, albeit in the context of somatic copy number alterations. ## Phasing with WhatsHap ### Preparing Input Files for WhatsHap WhatsHap requires two inputs: a VCF file with the offspring genotypes and a BAM file with the offspring reads. The parental genotypes are provided through a separate VCF file or through the pedigree information. The [Bioconductor project](https://bioconductor.org/) provides R packages for genomic data manipulation that can help you prepare and validate these input files. The key steps are as follows. First, ensure that the offspring VCF contains only high-quality variant calls. WhatsHap does not perform variant filtering, so you must do this before phasing. Second, create a pedigree file that specifies the family relationships. The pedigree file is a tab-separated text file with columns for family ID, individual ID, paternal ID, maternal ID, sex, and phenotype. For a trio, the pedigree file has three rows, one for each family member. The offspring row lists the father and mother IDs in the paternal and maternal columns. Third, index the BAM file and ensure that the sample name in the BAM file matches the sample name in the VCF file. WhatsHap uses the sample name to extract the correct reads from the BAM file. A mismatch between the BAM sample name and the VCF sample name will cause WhatsHap to fail or produce empty output. ### Running WhatsHap with Trio Mode The basic WhatsHap command for trio phasing is: ``` whatshap phase --ped pedigree.ped --reference reference.fasta offspring.vcf.gz offspring.bam -o phased.vcf.gz ``` WhatsHap uses the pedigree information to identify the parental samples and applies Mendelian constraints during phasing. The algorithm assigns each heterozygous variant in the offspring to a haplotype based on the combination of read evidence and parental genotypes. Variants that are homozygous in both parents provide the strongest phasing information because the offspring genotype at those positions is fully determined by Mendelian inheritance. The output of WhatsHap is a phased VCF file where each variant has a `PS` (phase set) tag and a `PID` (phase ID) tag. Variants with the same `PS` value belong to the same phase block. The `PID` tag indicates the physical position of the variant within the phase block. The phased genotypes are represented using the `|` separator, which indicates that the two alleles are phased with respect to each other. ### Phase Block Statistics After phasing, run `whatshap stats` to generate summary statistics for the phased VCF. The key metrics are the number of phase blocks, the phase block N50, and the proportion of heterozygous variants that are phased. The phase block N50 is the length such that 50 percent of all phased variants are in phase blocks of that length or longer. Higher N50 values indicate more contiguous phasing. For long-read data, you expect phase block N50 values in the hundreds of kilobases to megabases range. The [PEPPER-Margin-DeepVariant study](https://doi.org/10.1038/s41592-021-01299-w) reported that their pipeline produced phase blocks contiguously spanning between 85 and 92 percent of annotated genes across six samples. If your phase block N50 is substantially lower, this may indicate insufficient sequencing depth, low heterozygosity in the offspring, or problems with the variant calling step. ### Switch Error Rate Estimation The switch error rate is the proportion of adjacent phased variants that are assigned to the wrong haplotype relative to the true phase. Estimating the switch error rate requires a gold-standard phased dataset for comparison. For human samples, the Genome in a Bottle consortium provides benchmark calls for several well-characterized samples. For other organisms, you may need to generate your own validation data using an orthogonal method such as Strand-seq or linked-read sequencing. The [parent-of-origin detection study](https://doi.org/10.1016/j.xgen.2022.100233) reported an average mismatch error rate of 0.31 percent for single-nucleotide variants and 1.89 percent for insertions or deletions when using their methylation-based phasing approach. While this study used a different method, these error rates provide a useful reference point for the accuracy you might expect from long-read-based phasing. If your switch error rate is substantially higher, investigate whether the errors cluster in specific genomic regions, which may indicate alignment artifacts or variant calling errors. ## Quality Assessment and Validation ### Mendelian Consistency Checks Mendelian consistency is a fundamental quality check for trio data. At each variant position, the offspring genotype must be consistent with the parental genotypes under Mendelian inheritance. For example, if both parents are homozygous for the reference allele, the offspring must also be homozygous for the reference allele. If one parent is heterozygous and the other is homozygous for the reference allele, the offspring can be either heterozygous or homozygous for the alternate allele. Use a tool such as `bcftools +mendelian` or `rtg mendelian` to calculate Mendelian consistency statistics. The output reports the number of consistent and inconsistent sites, as well as the proportion of inconsistent sites. A Mendelian inconsistency rate above 1 percent suggests problems with variant calling, sample mix-ups, or pedigree errors. Investigate inconsistent sites to determine whether they cluster in specific genomic regions or have low quality scores. ### Sample Identity Verification Before proceeding with downstream analysis, verify that the three samples in your trio are actually related as expected. Sample mix-ups are a common source of errors in family-based studies. You can verify sample identity by comparing genotypes at highly polymorphic sites across the three samples. The [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/) provide access to databases of polymorphic markers that can be used for this purpose. A simple approach is to calculate the proportion of shared alleles between each pair of samples. Parents and offspring should share approximately 50 percent of their alleles at heterozygous sites. If the observed sharing is inconsistent with the expected relationships, investigate whether the samples were mislabeled during sequencing or data processing. ### Visual Inspection of Phased Regions While automated quality metrics are essential, visual inspection of phased regions can reveal problems that statistics miss. Use the Integrative Genomics Viewer (IGV) or a similar genome browser to examine regions of interest. Load the phased VCF, the offspring BAM file, and the parental BAM files. Check that the read alignments support the phased haplotypes and that the parental genotypes are consistent with the assigned phase. Pay particular attention to regions with low mapping quality, such as segmental duplications and low-complexity sequences. The [PEPPER-Margin-DeepVariant study](https://doi.org/10.1038/s41592-021-01299-w) showed that long-read methods can produce high-quality variants in these difficult regions, but errors are still possible. If you observe suspicious variant calls in these regions, consider masking them or applying additional quality filters. ## Reproducibility and Workflow Management ### Containerization and Version Control Reproducibility is a core requirement for published research. The [nf-core documentation](https://nf-co.re/docs) emphasizes the importance of containerization and version control for bioinformatics workflows. Use Docker or Singularity containers to pin the exact versions of all software tools in your pipeline. Record the container image identifiers and software versions in your project metadata. Version control with Git is essential for tracking changes to your analysis scripts and configuration files. The [The Carpentries Lessons](https://carpentries.org/lessons) provide training in Git and shell scripting that will help you establish good version control practices. Commit your scripts and configuration files to a repository at each major workflow stage, and tag releases that correspond to published results. ### Workflow Automation For complex analyses, consider using a workflow management system such as Snakemake or Nextflow. The [nf-core documentation](https://nf-co.re/docs) describes how to use Nextflow for reproducible pipeline execution. Workflow systems automatically track dependencies between steps, restart failed steps, and generate execution reports that document the exact commands and parameters used. The [Galaxy Training Network](https://training.galaxyproject.org/) provides tutorials on workflow construction that are accessible to researchers with limited programming experience. Galaxy also provides a graphical interface for running bioinformatics tools, which can be useful for exploratory analysis before you commit to a command-line workflow. ### Documentation Standards Maintain a laboratory notebook that records every analysis step, including the exact commands, software versions, and parameter settings. This notebook serves as the primary record of your analysis and is essential for reproducing results months or years later. The [EMBL-EBI Training portal](https://www.ebi.ac.uk/training) offers guidance on documentation standards for bioinformatics research. Include the following information in your notebook for each analysis run: the date, the input files and their checksums, the software versions, the command-line parameters, the output files, and any observations about unexpected results. This level of documentation may seem excessive, but it becomes invaluable when you need to troubleshoot problems or respond to reviewer requests. ## Common Failure Patterns and Troubleshooting ### Low Mapping Rates If the mapping rate for any sample is below 90 percent, investigate the cause before proceeding. Common causes include adapter contamination, sample degradation, or a mismatch between the sample and reference genome. Run a contamination check using a tool such as Kraken2 or Centrifuge to identify the taxonomic composition of your reads. If contamination is present, you may need to filter the reads or re-sequence the sample. ### Short Phase Blocks Short phase blocks can result from low sequencing depth, low heterozygosity, or errors in the variant calls. Check the depth and heterozygosity statistics for your offspring sample. If the depth is below 20x for nanopore data or 15x for PacBio HiFi data, consider sequencing additional data. If heterozygosity is low, which is common in inbred organisms or self-fertilizing species, you may need to use a different phasing strategy that relies less on read-based evidence. ### Mendelian Inconsistencies A high rate of Mendelian inconsistencies suggests problems with variant calling or sample identity. Check whether the inconsistencies cluster in specific genomic regions, which may indicate alignment artifacts. Verify that the sample names in the BAM and VCF files are correct and that the pedigree file accurately represents the family relationships. If the inconsistencies persist, consider using a stricter variant filtering threshold. ### WhatsHap Runtime Errors WhatsHap can fail for several reasons, including mismatched sample names, missing pedigree information, or malformed VCF files. Check the error message carefully and verify that the sample names in the BAM, VCF, and pedigree files are consistent. Ensure that the VCF file is properly compressed and indexed with `bgzip` and `tabix`. If WhatsHap reports that no reads support a variant, check that the BAM file contains reads from the correct sample. ## Limitations and Interpretation Boundaries ### Phasing Gaps in Low-Complexity Regions Even with long reads, some genomic regions remain difficult to phase. Low-complexity regions, such as homopolymer runs and tandem repeats, produce alignment errors that can break phase blocks. Segmental duplications and other highly repetitive regions may have ambiguous read placements that prevent accurate phasing. The [PEPPER-Margin-DeepVariant study](https://doi.org/10.1038/s41592-021-01299-w) demonstrated improved performance in these regions compared to short-read methods, but the limitations are not eliminated. When you encounter phasing gaps, document their genomic locations and consider whether they affect your downstream analysis. If a gene of interest falls within a phasing gap, you may need to use targeted approaches such as PCR-free long-range sequencing or optical mapping to resolve the phase. ### Interpretation of Phase Blocks A phase block represents a region where the relative phase of variants is known, but the absolute orientation, meaning which haplotype is maternal and which is paternal, is not determined by standard phasing. Trio phasing with WhatsHap assigns variants to haplotypes based on parental genotypes, but it does not label the haplotypes as maternal or paternal. If you need parent-of-origin information, you must use additional methods. The [parent-of-origin detection study](https://doi.org/10.1016/j.xgen.2022.100233) demonstrated that methylation-detecting nanopore sequencing combined with Strand-seq data can determine the parent of origin for chromosome-length haplotypes. This approach uses imprinted loci that are differentially methylated according to parent of origin. If parent-of-origin information is critical for your research, consider incorporating these methods into your workflow. ### Variant Calling Accuracy The accuracy of your phased haplotypes depends entirely on the accuracy of your variant calls. Errors in variant calling propagate to phasing errors, and these errors are difficult to detect after phasing is complete. The [PEPPER-Margin-DeepVariant study](https://doi.org/10.1038/s41592-021-01299-w) reported high accuracy for their pipeline, but the performance depends on sequencing depth, base quality, and the genomic context of each variant. For clinical applications, validate your phased variants using an orthogonal method before making any diagnostic conclusions. The [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/) provide access to reference materials and benchmark datasets that can be used for validation. The [EMBL-EBI Training portal](https://www.ebi.ac.uk/training) offers courses on variant interpretation that cover validation strategies. ## Professional Escalation Criteria ### When to Seek Expert Assistance Several situations warrant consultation with a bioinformatics specialist or the sequencing facility that generated your data. If you observe a mapping rate below 80 percent for any sample, this may indicate a problem with library preparation or sequencing that requires technical expertise to resolve. If your phase block N50 is more than 10-fold lower than expected for your sequencing platform, the problem may be in the data generation instead of the analysis. If you encounter persistent Mendelian inconsistencies that you cannot resolve through filtering or sample identity verification, consult with a genetic counselor or clinical geneticist before proceeding with any interpretation. Mendelian inconsistencies can indicate sample mix-ups, non-paternity, or rare genetic phenomena such as uniparental disomy, all of which have significant implications for research and clinical interpretation. ### Documentation for Publication When you publish results based on trio phasing, include detailed methods that describe your sequencing platform, coverage, variant calling approach, phasing software version, and quality metrics. The [nf-core documentation](https://nf-co.re/docs) provides guidance on reporting standards for bioinformatics analyses. Include your workflow scripts and configuration files in a public repository to enable others to reproduce your analysis. The [Galaxy Training Network](https://training.galaxyproject.org/) emphasizes the importance of reproducibility in bioinformatics education. Following these standards will make your work more credible and useful to the research community. The [Bioconductor project](https://bioconductor.org/) also provides resources for reproducible genomic analysis in R, which may be useful for downstream statistical analysis of your phased data. ## Frequently Asked Questions ### What is the minimum sequencing depth required for trio phasing with long reads? For nanopore data, a minimum of 20x coverage for the offspring sample is recommended for reliable variant calling and phasing. For PacBio HiFi data, 15x coverage is often sufficient because of the higher per-read accuracy. Parental samples can be sequenced at lower depth, approximately 10x, because they are used primarily for genotype comparison instead of read-based phasing. However, higher depth in parents improves the accuracy of genotype calls at heterozygous positions, which strengthens the Mendelian constraints used by WhatsHap. ### Can I use short-read data from parents with long-read data from the offspring? Yes, this hybrid approach is valid and can be cost-effective. The parental short-read data provides accurate genotypes at most variant positions, while the offspring long-read data provides the read-level evidence needed for phasing. WhatsHap supports this configuration by accepting a BAM file from the offspring and VCF files from the parents. The main limitation is that parental genotypes at low-complexity regions may be less accurate with short reads, which can reduce phasing accuracy in those regions. ### How do I determine whether two variants are in cis or in trans? Two variants are in cis if they are on the same chromosome copy and in trans if they are on opposite copies. After phasing, you can determine the relationship by examining the phased genotypes in the VCF file. If the two variants have the same phase set and the alternate alleles are on the same haplotype, they are in cis. If the alternate alleles are on different haplotypes, they are in trans. This information is critical for interpreting compound heterozygosity in recessive diseases. ### What is the difference between a phase block and a haplotype? A phase block is a genomic region where the relative phase of variants is known with confidence. Within a phase block, all variants are assigned to one of two haplotypes. A haplotype is the complete set of alleles on a single chromosome copy. A phase block is a contiguous segment of a haplotype, but the entire haplotype may consist of multiple phase blocks separated by regions where phasing was not possible. The phase block N50 statistic summarizes the contiguity of phasing across the genome. ### How can I improve phase block contiguity in regions of low heterozygosity? Low heterozygosity reduces the density of informative variants that can be phased. To improve contiguity, you can increase sequencing depth to detect more heterozygous variants, use a more sensitive variant caller to recover low-quality variants, or incorporate additional data sources such as linked-read sequencing or optical mapping. For some organisms, you can also use population-based phasing approaches that leverage linkage disequilibrium information from many individuals. ### What should I do if WhatsHap produces no phased variants? If WhatsHap produces no phased variants, check that the offspring VCF contains heterozygous variants and that the BAM file contains reads from the correct sample. Verify that the sample names in the BAM, VCF, and pedigree files are consistent. Check that the pedigree file correctly specifies the family relationships and that the parental samples are present in the VCF file. If the VCF file contains only homozygous variants, WhatsHap has nothing to phase, which may indicate a problem with variant calling or sample identity. ### Can I phase structural variants with WhatsHap? WhatsHap is designed for small variants, primarily single-nucleotide variants and small insertions or deletions. Structural variants are typically phased indirectly by using the phase information from nearby small variants. If a structural variant overlaps a phased small variant, you can infer the haplotype assignment of the structural variant. Dedicated structural variant phasing tools are under development, and the [haplotype-aware segmentation study](https://doi.org/10.1186/s13059-026-03971-w) demonstrates the broader utility of haplotype information for structural analysis. ### How do I report phasing results in a publication? Report the sequencing platform, coverage, read length distribution, variant calling method, phasing software version, and key quality metrics including phase block N50 and switch error rate. Describe the reference genome version and the filtering criteria applied to the variant calls. Include the workflow scripts and configuration files in a public repository. The [nf-core documentation](https://nf-co.re/docs) provides guidance on reporting standards, and the [EMBL-EBI Training portal](https://www.ebi.ac.uk/training) offers courses on data sharing and publication practices. ## Related Bioinformatics Guides - [De Novo Genome Assembly with Long Reads: A Practical Workflow](/knowledge/bioinformatics/de-novo-genome-assembly-with-long-reads-a-practical-workflow) - [RNA-Seq Data Analysis Workflow: From Raw Reads to Insights](/knowledge/bioinformatics/rna-seq-data-analysis-workflow-from-raw-reads-to-insights) - [RNA Sequencing Data Analysis: From Raw Reads to Differential Expression](/knowledge/bioinformatics/rna-sequencing-data-analysis-from-raw-reads-to-differential-expression) - [Single-Cell Sequencing Workflow: From Sample Preparation to Data Analysis](/knowledge/bioinformatics/single-cell-sequencing-workflow-from-sample-preparation-to-data-analysis) - [Full-Length Transcript Sequencing: Unraveling Isoform Diversity with Long Reads](/knowledge/bioinformatics/full-length-transcript-sequencing-unraveling-isoform-diversity-with-long-reads) ## References and Further Reading - [NCBI Data Resources](https://www.ncbi.nlm.nih.gov/). National Center for Biotechnology Information. - [EMBL-EBI Training](https://www.ebi.ac.uk/training). European Bioinformatics Institute. - [Bioconductor](https://bioconductor.org/). Bioconductor Project. - [Galaxy Training Network](https://training.galaxyproject.org/). Galaxy Project. - [nf-core Documentation](https://nf-co.re/docs). nf-core. - [The Carpentries Lessons](https://carpentries.org/lessons). The Carpentries. - [Haplotype-aware segmentation with HapASeg increases accuracy of detecting homolog-specific somatic copy number alterations.](https://doi.org/10.1186/s13059-026-03971-w). 2026. - [Scalable, accessible and reproducible reference genome assembly and evaluation in Galaxy.](https://doi.org/10.1038/s41587-023-02100-3). 2024. - [Extensive variation between chromosomes of North American and European hop.](https://doi.org/10.1038/s41467-026-72379-8). 2026. - [Parent-of-origin detection and chromosome-scale haplotyping using long-read DNA methylation sequencing and Strand-seq.](https://doi.org/10.1016/j.xgen.2022.100233). 2023. - [Haplotype-aware variant calling with PEPPER-Margin-DeepVariant enables high accuracy in nanopore long-reads.](https://doi.org/10.1038/s41592-021-01299-w). 2021. > This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.