How to Find Your Haplogroup From Raw Data

Finding your haplogroup from raw genetic data typically involves uploading a file you downloaded from a consumer DNA testing company to a free online classifier tool. The specific tool depends on which haplogroup you want: your maternal (mitochondrial DNA) haplogroup or your paternal (Y-chromosome) haplogroup. The process is straightforward for most people, but the accuracy of your result depends heavily on which tool you pick and what format your data is in.

Two Haplogroups, Two Separate Processes

Everyone inherits mitochondrial DNA (mtDNA) from their mother, which means everyone can determine a maternal haplogroup. The Y chromosome, by contrast, passes strictly from father to son without recombining during reproduction, so only people who carry a Y chromosome can determine a paternal haplogroup.¹ These two lineages are analyzed with completely different tools using different parts of your raw data, so you will go through a separate process for each one.

If you tested with a company like 23andMe, AncestryDNA, or MyHeritage, your raw data file contains genotype calls at hundreds of thousands of positions across your genome. Buried in that file are the mitochondrial and (if applicable) Y-chromosome markers that classifiers use to assign you to a branch on the human family tree. The challenge is that different companies use different chip versions, sequence different markers, and export data in slightly different formats.² Your first job is figuring out what your file actually contains and which tools can read it.

Finding Your Maternal Haplogroup

The most widely used tool for mtDNA haplogroup classification is Haplogrep, now in its third version. It is a free web application where you upload your mitochondrial sequence data, and it returns your haplogroup assignment along with a quality score indicating how confident the classification is. Haplogrep works with FASTA files (raw sequence text), VCF files (variant call lists), and even partial sequences covering just the mitochondrial control region.³ The original version was designed to handle thousands of mtDNA profiles at once and has been used extensively in population genetics and forensic work.⁴

If you tested through a consumer company, your raw data is typically a text file with genotype calls at specific positions, not a full mitochondrial sequence. Some tools handle this directly, but others expect a FASTA or VCF file. Free conversion utilities exist online that can reformat consumer genotype data into something Haplogrep accepts. If your testing company already reported a haplogroup, running it through an independent tool is still worthwhile as a check, especially if you want a more granular sub-haplogroup assignment than the company provided.

There are other mtDNA classifiers besides Haplogrep. James Lick’s mtHap is a popular web-based option in the genetic genealogy community. EMMA and Haplocheck are additional tools that have performed well in accuracy testing. The landscape of available classifiers is broad enough that a benchmarking study evaluated eleven of them to see how they compared.

How Accurate Are mtDNA Classifiers?

The accuracy gap between the best and worst mtDNA haplogroup tools is enormous. When researchers benchmarked classifiers against whole-genome sequencing data, Haplogrep, Haplocheck, EMMA, and James Lick’s mtHap all classified every sample correctly. At the other end, MixEmt correctly classified only about 31% of the analyzed samples from the same whole-genome data.⁵ That is a staggering difference, and it means picking the wrong tool can give you a completely wrong haplogroup assignment with no warning.

Performance also depends on what type of data you feed the tool. When the same benchmarking study tested tools on whole-exome sequencing data instead of whole-genome data, accuracy dropped for most classifiers. Only Haplocheck classified all haplogroups correctly from exome data, and only when provided a BAM file rather than other formats.⁵ Consumer genotyping data is sparser than either of these, covering fewer mitochondrial positions, so results can be less precise. You might get an accurate assignment to a major haplogroup branch but miss the finer sub-branches that a full mitochondrial sequence would resolve.

Finding Your Paternal Haplogroup

Y-chromosome haplogroup classification follows a similar logic but uses different tools and different markers. The Y chromosome does not recombine across most of its length, making it a clean tracer of paternal lineage through specific mutations called Y-SNPs.¹ Y-SNPs are the gold standard for haplogroup assignment, though some forensic and genealogy applications also use Y-STRs (short tandem repeats), which are less precise for haplogroup inference.⁶

Yleaf is a freely available software tool designed specifically for Y-chromosome haplogroup inference from next-generation sequencing data. It works regardless of the library preparation or sequencing method used, which makes it flexible across different data types.⁷ A newer tool called yallHap integrates a much larger reference tree of over 185,000 Y-SNPs and uses probabilistic scoring to assign haplogroups. Validation on over 1,200 high-coverage genomes achieved about 99.9% accuracy.⁸

For people working with consumer genotyping data rather than full sequencing, the situation is trickier. Consumer chips test only a fraction of Y-chromosome SNPs, so the depth of your haplogroup assignment will be limited. You might learn you belong to haplogroup R1b, for example, but not which of its many downstream branches you sit on. Services like FTDNA’s Y-DNA tests sequence more Y-chromosome markers specifically, which allows deeper classification. If you already have consumer raw data and want to go deeper, a dedicated Y-DNA test from a company that specializes in it is the usual next step.

Getting Your Raw Data Into the Right Format

The most common stumbling block is format mismatch. Consumer companies export your genotype data as a zipped text file, usually tab-delimited, with columns for chromosome, position, and the genotype call at that position. But many classification tools expect BAM files (aligned sequencing reads), VCF files (variant calls against a reference), or FASTA files (raw sequence text).⁵ These are fundamentally different data types.

If you have whole-genome sequencing data from a service like Nebula Genomics or Dante Labs, you likely already have BAM and VCF files, which most tools accept natively. If you only have consumer chip data from 23andMe or AncestryDNA, you need a tool that either reads that format directly or a converter that extracts the mitochondrial or Y-chromosome markers and reformats them. Several community-built tools in the genetic genealogy space handle this conversion, though you should verify that the conversion preserves all the markers your classifier needs.

One subtle issue worth knowing about is the reference genome version your data is aligned to. Older data may use GRCh37, while newer data uses GRCh38. Some newer tools also support T2T-CHM13, the telomere-to-telomere assembly.⁸ If your data is aligned to one reference genome and the tool expects another, positions will not match up correctly, and you will get wrong results. Most modern tools handle multiple reference versions, but older or simpler tools may not.

Why Different Tools Give Different Answers

If you run the same raw data through multiple classifiers, you may get slightly different haplogroup assignments. This is not necessarily a sign that something is broken. The reference phylogenetic tree that each tool uses matters enormously. These trees are updated regularly as researchers discover new mutations and refine branch points. A tool using an older tree version will assign you to broader, less specific branches. A tool using the latest version with nearly 186,000 Y-SNPs will resolve your lineage much more finely.

The scoring method also differs between tools. Some use a simple rule-based approach, checking which mutations you carry and matching them against the tree. Others use probabilistic scoring that accounts for missing data, sequencing errors, or ambiguous calls. Probabilistic methods tend to perform better when your data is incomplete, which is exactly the situation with consumer genotyping chips that only test a subset of relevant markers.

For mtDNA specifically, how a tool handles insertions and deletions in the mitochondrial sequence affects accuracy. Haplogrep 3 addressed this by implementing alignment rules that correct common problems with indel calling, which was a known shortcoming of earlier versions.³ If you get an unexpected result from one tool, running your data through a second well-validated tool is a reasonable sanity check.

Privacy Risks of Uploading Your Data

Before you upload your raw genetic data to any third-party tool, consider what you are sharing. Your genotype file contains far more than haplogroup markers. It typically includes hundreds of thousands of SNPs across your entire genome, and researchers have demonstrated that SNP data can be used to re-identify supposedly anonymous individuals. In one well-known case, uploading about 700,000 SNPs to a genetic genealogy database was enough to identify a research participant by surname through matches with relatives.⁹

Web-based tools that run in your browser without uploading data to a server are the safest option from a privacy standpoint. Some tools do process data server-side, and their privacy policies vary. A few practical steps can reduce your risk:

• Strip your file: If you only need your haplogroup, extract just the mitochondrial or Y-chromosome markers from your raw data before uploading. This avoids exposing autosomal SNPs that could be used for re-identification.
• Use local tools: Command-line tools like Yleaf or yallHap run on your own computer, so your data never leaves your machine.
• Read the terms: Some web platforms retain uploaded data or use it for research. Check whether you can delete your data after getting your result.

Running a local tool requires some comfort with the command line, but the privacy benefit is substantial if you are cautious about your genetic data.

What Consumer Chips Miss

Consumer genotyping chips were designed primarily for health and ancestry estimates across the autosomal genome, not for deep haplogroup resolution. They test a limited set of mitochondrial and Y-chromosome markers compared to what dedicated sequencing covers. This means the haplogroup you get from consumer raw data is often a broad branch assignment rather than a fine-grained terminal subclade.

Mitochondrial DNA is relatively small at around 16,500 base pairs, and about 2.4% of that genome shows common variation between individuals.¹⁰ Consumer chips capture some of these variable positions, but full mitochondrial sequencing captures all of them. For maternal haplogroup assignment, this difference can mean landing on a major branch like H or K from chip data versus landing on a specific sub-branch like H1a1 from full sequencing.

For Y-DNA, the gap is even wider. The non-recombining region of the Y chromosome spans millions of base pairs, and consumer chips test only a small fraction of the known Y-SNPs. Dedicated Y-DNA tests, like those offered by FamilyTreeDNA, sequence far more of the relevant region. If your consumer-chip haplogroup assignment seems vague or you want to compare your lineage at a fine-grained level with other people in genealogy databases, upgrading to a dedicated Y-DNA test is the practical solution.

When Your Haplogroup Has Medical Relevance

Haplogroups are mostly discussed in the context of ancestry and genealogy, but a growing body of research links specific haplogroups to disease susceptibility. These associations are generally modest in effect size and population-specific, but they are real.

In a study of Danish patients, mitochondrial haplogroup H was found at a higher frequency in people with hypertrophic cardiomyopathy compared to controls (60% versus roughly 41–46%), while haplogroup J appeared protective.¹¹ Separately, among Black veterans with HIV, mitochondrial haplogroup L3 was associated with a higher risk of developing diabetes, with an adjusted hazard ratio of about 1.56.¹²

These findings are interesting but come with strong caveats. Haplogroup-disease associations tend to be small relative to other risk factors like lifestyle, family history of specific conditions, or individual genetic variants elsewhere in your genome. They also tend to be population-specific: a protective haplogroup in a Danish study may not have the same effect in an East Asian or West African population. Knowing your haplogroup is not a substitute for clinical genetic testing if you are concerned about a specific disease risk.

Ancient DNA and the Expanding Tree

One reason the haplogroup classification landscape keeps changing is that the reference trees are constantly being revised. Every time researchers sequence ancient DNA from archaeological remains, they potentially discover new branches or confirm the age and geography of existing ones. Classifiers like Haplotracker have been used to assign mitochondrial haplogroups to 2,000-year-old skeletal remains from elite Xiongnu burials in Mongolia.¹³ Tools like yallHap now include built-in filtering for the kinds of DNA damage patterns common in ancient samples, which cause specific error types that modern-sample classifiers are not designed to handle.⁸

For the average person running their consumer data through a classifier, the practical implication is simple: your haplogroup assignment might become more specific over time as the trees grow. If you ran your data through Haplogrep three years ago and got “H,” running it again through the latest version with its updated tree might refine that to “H1a” or even deeper. Revisiting your classification every year or two is not a bad idea, especially if you have whole-genome or whole-mitochondrial sequencing data that contains the markers needed for finer resolution.

A Step-by-Step Walkthrough

If you have never done this before, here is the practical sequence for getting your haplogroup from consumer raw data:

• Download your raw data: Log into your testing company’s website and look for a “download raw data” option, usually in account settings. The file will typically be a zipped text file.
• Identify what you have: Open the file in a text editor. You will see rows of genotype calls with chromosome numbers. Look for entries on the mitochondrial chromosome (often labeled “MT” or “26”) and, if applicable, the Y chromosome.
• Choose your tool: For maternal haplogroup, Haplogrep 3 is the most validated web-based option. For paternal haplogroup from consumer data, community tools on genetic genealogy forums often handle the conversion and classification in one step.
• Convert if needed: If your chosen tool does not accept raw consumer format, use a converter to extract the relevant markers into a VCF or FASTA file.
• Upload and classify: Run the tool and check the quality score. A high-confidence assignment with a good quality score is reliable. A low-confidence one may reflect missing markers rather than an error.
• Cross-check: Run a second tool if you want extra confidence. If two well-validated tools agree, the assignment is almost certainly correct.

The whole process takes about fifteen minutes once you know what you are doing. The first time, expect to spend some time figuring out your file format and finding the right tool, but there is no specialized knowledge required beyond basic comfort with downloading files and using a web interface.

Heteroplasmy and Other Complications

Mitochondrial DNA has a quirk that can complicate haplogroup assignment: heteroplasmy. Unlike nuclear DNA, where you have two copies of each chromosome, you have thousands of copies of your mitochondrial genome in each cell, and not all of them are always identical. Research on mother-offspring pairs has found that roughly 45% of individuals carry a mixed population of mitochondrial DNA.¹⁰ Most of the time, heteroplasmy does not affect your haplogroup assignment because the variants involved are minor and the dominant sequence is clear. But in rare cases, a heteroplasmic site can fall on a branch-defining mutation, creating ambiguity about which branch you belong to. Specialized tools like Haplocheck were designed partly to detect and account for this.¹⁴

Consumer genotyping chips generally will not detect heteroplasmy at all, because they report a single genotype call at each position. Full mitochondrial sequencing, especially at high depth, is needed to identify heteroplasmic sites. If your haplogroup assignment from consumer data seems inconsistent with your known maternal ancestry, heteroplasmy is one possible explanation, though a more common reason is simply that the chip missed key markers.