Learn Explainer
What a haplogroup is: one line of descent among thousands
A haplogroup names one branch of a family tree drawn from DNA, following the father's father's line or the mother's mother's back tens of thousands of years. Ten generations back there are 1,024 places in your family tree, and a haplogroup follows one of them.
Key takeaways
- A haplogroup is a branch of the Y-chromosome or mitochondrial tree, defined by mutations everyone on the branch shares. The Y follows the father's father's line, mitochondrial DNA the mother's mother's[1, 2].
- Each line follows one ancestor per generation. By our arithmetic, ten generations back the two lines follow 2 of 1,024 ancestral places, and thirty generations back 2 of more than a billion.
- Names change because the trees grow. The Y tree held 153 branches in 2002[8] and 69,676 in YFull's September 2026 release[10], and the branch most Finnish men carry has been called N1c, N3 and N1a1[16, 18, 19].
- A chip reads a fixed set of positions and can only reach the branches they define. A sequenced genome reads far more, so the same person can get a longer name from it.
- Aimosti places the maternal line on PhyloTree Build 17 with Haplogrep 3 and reads 13 Y markers for the paternal line; on a sparse chip it names only the branch the evidence agrees on.
Two parts of the genome are passed down without being shuffled. Mitochondrial DNA goes from a mother to all her children[1], and the male-specific part of the Y chromosome, 95 percent of its length, goes from father to son without recombining[2]. A new mutation on either is copied to every descendant down that line. Sorting people by the mutations they share gives a tree, and a haplogroup is one branch of it: the people who descend down one line from the person in whom a particular mutation first appeared.
Two lines that skip the reshuffle
Most of your genome is a blend. Each of the 22 numbered chromosome pairs is cut and recombined before it is passed on, so a grandparent's contribution arrives in pieces and thins with every generation. Any Neanderthal DNA a person carries sits in this blended part, as scattered pieces. Two small parts of the genome do not work like that.
Mitochondria, small structures inside our cells, carry their own DNA: a ring of 16,569 bases in the reference sequence[3]. A child inherits it from the mother. The first family study to show this, in 1980, followed a variant through three generations of a 33-member family, and every child carried the mother's pattern[1]. Mitochondrial DNA does not recombine, so it reaches each generation whole[4].
The Y chromosome passes from father to son. Its male-specific region, which makes up 95 percent of its length, has no partner to swap pieces with[2]; only the short regions at its tips, which pair with the X chromosome, recombine[5]. A man's Y is his father's Y, give or take a few new mutations.
Those new mutations are what make the lines readable. Helgason and colleagues measured the rate in 274 Icelandic paternal lines at 8.71 × 10⁻¹⁰ mutations per position per year, across 15.2 million positions of the Y[6]. At the average age of fathers estimated by Wang and colleagues, 30.7 years[7], that comes to about 0.4 new mutations per generation in those positions, by our own estimate: roughly one every two or three generations. Each one stays with every son and grandson after it, marking them all as one branch.
Terms this guide relies on
- Haplogroup
- A branch of the Y or mitochondrial tree: everyone who carries the mutations that define it. A smaller branch inside it is a subclade.
- Marker
- A position whose changed letter defines a branch. Y markers have names such as M170 or M253; mitochondrial ones are written as the position and the new letter.
- Derived and ancestral
- At a marker, the derived letter is the changed one, carried by the branch; the ancestral letter is the one before the change.
Trees drawn from shared mutations
Researchers sequence or genotype many Y chromosomes or mitochondrial genomes, find the mutations they share, and arrange them into the tree that explains the pattern with the fewest changes. The Y Chromosome Consortium did this in 2002 with 245 markers and found 153 branches[8]. The Consortium also set rules for naming the branches, so that each name says where on the tree the branch sits.
16,569
bases in the reference mitochondrial genome, short enough to read end to end[3]
over 5,400
branches in PhyloTree Build 17, the mitochondrial tree, last updated in February 2016[9]
69,676
branches in YFull's Y-chromosome tree, release 14.06.00, September 2026[10]
The mitochondrial tree most software uses is PhyloTree, compiled by van Oven and Kayser and first published in 2009[11]. Its last build, number 17, dates from 18 February 2016 and lists over 5,400 branches[9]. The Y tree has two main public versions: the one kept by the International Society of Genetic Genealogy (ISOGG), and YFull, which builds its tree from sequenced Y chromosomes and gives every branch an estimated age[10].
| Y chromosome | Mitochondrial DNA | |
|---|---|---|
| Passed down | Father to son | Mother to every child |
| Recombination | None across the male-specific region, 95% of the chromosome | None |
| Length in the reference | 57,227,415 bases (GRCh38) | 16,569 bases (the rCRS) |
| Readable by short-read sequencing | About 10 million positions, once repeats are filtered out | The whole genome |
| Main public trees | ISOGG; YFull, 69,676 branches | PhyloTree Build 17, over 5,400 branches |
| How branches are named | Letters and digits (N1a1), or the letter and a marker (N-M46) | Letters and digits (H1a, B4b1c2) |
Source: NCBI RefSeq; Skaletsky et al. 2003; Kivisild 2015; Poznik et al. 2013; PhyloTree; YFull[2, 3, 4, 5, 9, 10, 12].
The size gap shapes what a file can show. The mitochondrial genome is short enough to read end to end. Much of the Y is repeats that short reads cannot place: Poznik and colleagues, sequencing 69 men, kept 9.99 million of its 57 million positions and found 11,640 single-letter variants there[5].
Placing a sample on the tree
Placement walks the tree from the root and asks, branch by branch, whether the sample carries the mutations that define it. A mitochondrial genome carries dozens of markers along its path, so software scores every branch. Haplogrep, from the Medical University of Innsbruck, is the standard tool[13].
How Haplogrep places a mitochondrial genome
- List where the sample differs from the reference sequence, the rCRS. For VCF input, Haplogrep 3 first runs a quality check[13].
- Walk the tree, and for every branch compare the mutations expected on the path to it with the ones the sample has.
- Score each branch with the Kulczynski measure: the average of two shares, the branch's expected mutations that the sample has and the sample's mutations that the branch expects, with each mutation given its own weight[14].
- Rank the branches and report the best one with its quality score, alongside the next best hits, the expected mutations and the ones left over[13].
Leftover mutations are information too: private mutations nobody has put on the tree, sequencing errors, or a mixed sample, which Haplogrep 2 checks for and lists as potentially contaminated[15]. With genotyping-chip data, Haplogrep 3 narrows the comparison to the positions the chip measured[13].
Y placement follows the same logic. A test that reads a fixed list of markers, as our report does, can place a man no deeper than the list reaches.
Why the names change
The 2002 Consortium's names are hierarchical[8]. A label such as N1c1a1a, used by Lamnidis and colleagues for one branch[16], reads as a path: haplogroup N, its branch 1, that branch's branch c, and so on down. That makes the name a map reference, and a map reference changes when the map does. A new split found above a known branch pushes every label below it one step along.
| Year | Tree | Branches | Markers |
|---|---|---|---|
| 2002 | Y Chromosome Consortium | 153 | 245 genotyped |
| 2008 | Karafet and colleagues, a revision of the 2002 tree | 311 | About 600 |
| 2026 | YFull, release 14.06.00 | 69,676 | 459,477 SNPs |
Source: Y Chromosome Consortium 2002; Karafet et al. 2008; YFull tree statistics, release 14.06.00 of 22 September 2026[8, 17, 10].
The branch most Finnish men carry shows how much a name can move. It is defined by a marker called M46, also known as Tat. Lamnidis and colleagues called it N1c in 2018[16], Ilumäe and colleagues called it N3[18], and Preussner and colleagues, who compared their branches with ISOGG's tree version 15.73, called it N1a1 in 2025, noting the older labels N1c1 and N3[19]. YFull lists it as N-TAT[10]. Four labels, one set of men. Our guide to Finnish DNA traces that branch and its two Finnish sub-branches in detail.
Names built from a marker avoid this. N-M46 means the branch defined by M46, wherever the tree places it. Preussner and colleagues named their branches that way, after YFull, and listed each one's other names in their supplementary tables[19]. When a report, a forum and a paper seem to disagree about a haplogroup, the marker is the thing to compare.
The mitochondrial tree changes more slowly. The most up-to-date one the Haplogrep authors cite, built on PhyloTree, holds 6,401 branches against Build 17's 5,400 or so[13, 9], and the same genome can get a longer name on the larger tree. A placement means little without the name of the tree it was read against.
One line among thousands
Every person has two parents, four grandparents and eight great-grandparents. The number of places in a family tree doubles with each generation back, and a haplogroup follows one of them.
The arithmetic is our own, and the years are estimates. Wang and colleagues put the average human generation at 26.9 years[7]. Ten generations, about 270 years, give 1,024 ancestral places. Twenty give 1,048,576. Thirty generations, about 800 years, give 1,073,741,824 places, more than twice the 450 million people the Population Reference Bureau estimates were alive in the year 1200[20]. Forty give more than a trillion, about nine times its estimate of everyone ever born, 117 billion[20].
Those places cannot all be different people; the same ancestors fill many of them, which genealogists call pedigree collapse. The point survives it. A few hundred years back, the two lines follow two people out of a crowd, and the rest of the crowd is invisible to both.
The lines are long. Every Y chromosome alive today goes back to one man, and every mitochondrial genome to one woman. Poznik and colleagues, applying the same method to both, estimated that the man lived 120,000 to 156,000 years ago and the woman 99,000 to 148,000 years ago[5]. Every haplogroup is a branch somewhere between those roots and the present.
What a haplogroup does tell you is narrow and real. It places one line on a dated tree, and in research, where thousands of lines are counted together, it records migrations and expansions: Poznik and colleagues read bursts of male population growth in each of five continental groups from 1,244 Y chromosomes[21].
Chips and genomes read the tree to different depths
A genotyping chip measures a fixed list of positions chosen when the array was designed, and which mitochondrial and Y positions are on it depends on the array[22]. What is not on the list cannot be read. A sequenced genome reads nearly all of the mitochondrial genome and the readable part of the Y, so a branch defined by a rare marker is within reach.
Sparse chips need care for a less obvious reason. When only a few measured positions differ from the reference, one of them can decide the answer alone, and a single wrong call can move the placement to another branch with a high quality score. The score measures how well the sample fits the branch it was given; it cannot see the positions the chip never measured.
Our report handles this in two ways. When several branches score within 0.01 of the best one, it names the branch they share instead of picking one. And when a chip's best placement rests on fewer than 7 of the expected markers, it places the line again once for each derived call, leaving that call out each time, and reports only the branch every answer has in common. If too few of those answers hold up, it leaves the line unplaced and says why.
The Y has an extra problem on chips. Our report finds its 13 Y markers by position alone, and a position means something only once the genome build of the export is known. 23andMe and AncestryDNA exports give the build; MyHeritage and FamilyTreeDNA exports do not, so for those the report says the paternal line could not be read instead of guessing[22]. MyHeritage kits sold since October 2025 are sequenced, and their raw-data export leaves the Y chromosome out entirely[23].
What the Aimosti report shows
Each line gets a card. A placed maternal line shows the haplogroup, its path down the tree, the placement quality with how many expected markers were found, whether it was read from a chip (with the number of positions measured) or a full mitochondrial sequence, and the tool and tree it was read against. A paternal line shows its name, its path and how many of its diagnostic markers were present.
| Chip export | Plain VCF | gVCF | BAM or CRAM | |
|---|---|---|---|---|
| Maternal line | Haplogrep 3 on PhyloTree Build 17, against the positions the chip measured | The whole tree; a position with no record is taken to match the reference | As a plain VCF | From the variant file uploaded with it; the reads' mitochondrion is not fetched |
| Quality needed | 0.80 plainly, 0.70 with a caveat | 0.90 plainly, 0.80 with a caveat | 0.90 plainly, 0.80 with a caveat | As for the variant file |
| Paternal line | 13 markers, from 23andMe and AncestryDNA exports; MyHeritage and FamilyTreeDNA exports are reported as unreadable | 13 markers; one counts only when the file lists its derived letter | As a plain VCF | From the variant file uploaded with it |
| Deepest Y branch | Broad branches such as N, I1 or R1a | The same | The same | As for the variant file |
Source: Aimosti's report as of October 2026; the coverage and supported-files pages list every module by file type[22, 23].
The two lines are read by different machinery on purpose. The maternal line has a full tree and a scoring tool, so a sequenced file is usually placed several levels deep, on a branch such as H1a or B4b1c2, and a branch we have not yet written about is still placed and named; our written notes cover 30 major mitochondrial branches[22]. A mitochondrial call whose allele fraction, the share of the reads carrying it, is below 0.90, or below 0.50 for an insertion or deletion, is left out, because a low-fraction call is more often noise than lineage.
The paternal line is read from 13 markers naming 13 broad branches: A, B, C, E, G, I, I1, I2, J2, N, O, Q and R1a. The report names the deepest one whose marker is derived. The reference Y is itself an R1b sequence, so R1b has no marker here: a man whose file carries a Y but none of the 13 derived letters is told this most often means R1b, and the line is left unnamed, because a match to the reference names no branch. A file with no Y data gets a note in place of a name, worded by what the file shows: an export with no Y positions at all is told so, and a chip export whose build could not be established is told that instead.
The report also estimates an ancestry composition from the rest of the genome, against a fixed reference panel. It is a different measurement, and the two can point in different directions without either being wrong; why those percentages differ between companies has a guide of its own. The free file check reads a file's format and genome build in your browser before anything is uploaded.
What Aimosti would (and wouldn't) show you
The report places the maternal line with Haplogrep 3 on the whole of PhyloTree Build 17, from a chip or a sequenced file, with the placement quality and the markers found. It reads the paternal line from 13 Y-chromosome markers, which name a broad branch such as N, I1 or R1a. A BAM or CRAM reaches both lines through the variant file uploaded with it.
What we won't claim
A haplogroup is one maternal and one paternal line among thousands of ancestral lines. We won't turn it into an ethnicity, a nationality, a race or a percentage, name a branch deeper than the file's markers support, or present a reference-matching Y as a confirmed lineage.
Bottom line. A haplogroup answers a narrow question precisely: which branch of the Y or mitochondrial tree one line of descent belongs to. Its name depends on the tree it was read against and its depth on how much the file can see.
Questions people ask
Can a woman find her paternal haplogroup in her own DNA file?
No. The Y chromosome passes only from father to son[2], so a woman's file has no Y to read. Her father's paternal line is carried by her father, his brothers and their sons. Her own file does carry her maternal line, which follows her mother's mother back.
Is haplogroup N1c the same as N1a1?
Yes. Both name the Y branch defined by the marker M46, also called Tat. Lamnidis and colleagues used N1c in 2018[16], Ilumäe and colleagues N3[18], and Preussner and colleagues, who compared their branches with ISOGG's tree version 15.73, N1a1; they list N1c1 and N3 as its other names[19]. YFull calls it N-TAT[10].
Was mitochondrial Eve the only woman alive in her time?
No. She is the most recent woman from whom every living person's mitochondrial DNA descends, dated to 99,000 to 148,000 years ago in one study[5]. Other women of her time can be ancestors of people today through the rest of the genome; their mitochondrial lines did not survive. The same holds on the Y side.
Why does my chip give a shorter haplogroup name than a whole genome?
A chip reads only the positions on its array, and Haplogrep scores it only against those[13], so it can place a line no deeper than its markers reach. A whole genome reads nearly every mitochondrial position, so the same maternal line usually lands several branches deeper.
References
- Giles RE, Blanc H, Cann HM, Wallace DC. Maternal inheritance of human mitochondrial DNA. Proceedings of the National Academy of Sciences of the USA, 1980. doi:10.1073/pnas.77.11.6715 A three-generation family of 33 members; in each family where the parents differed, the progeny carried the maternal pattern (abstract).
- Skaletsky H, Kuroda-Kawaguchi T, Minx PJ, et al. The male-specific region of the human Y chromosome is a mosaic of discrete sequence classes. Nature, 2003. doi:10.1038/nature01722 The MSY 'differentiates the sexes and comprises 95% of the chromosome's length' (abstract).
- Homo sapiens mitochondrion, complete genome (NC_012920.1, the revised Cambridge Reference Sequence). NCBI Nucleotide, 2026. Sequence length 16,569 bp, read from NCBI E-utilities esummary on 10 October 2026.
- Kivisild T. Maternal ancestry and population history from whole mitochondrial genomes. Investigative Genetics, 2015. doi:10.1186/s13323-015-0022-2
- Poznik GD, Henn BM, Yee MC, et al. Sequencing Y chromosomes resolves discrepancy in time to common ancestor of males versus females. Science, 2013. doi:10.1126/science.1237619 69 men; Y filtered to 9.99 million sites, 11,640 SNVs; TMRCA of the Y 120 to 156 ky and of mtDNA 99 to 148 ky.
- Helgason A, Einarsson AW, Guðmundsdóttir VB, et al. The Y-chromosome point mutation rate in humans. Nature Genetics, 2015. doi:10.1038/ng.3171 753 Icelandic males in 274 patrilines; 8.71 × 10⁻¹⁰ mutations per position per year over 15.2 Mb of XDG, XTR and rAMP sequence (abstract).
- Wang RJ, Al-Saffar SI, Rogers J, Hahn MW. Human generation times across the past 250,000 years. Science Advances, 2023. doi:10.1126/sciadv.abm7047 Average generation time 26.9 years; fathers 30.7, mothers 23.2 (abstract).
- The Y Chromosome Consortium. A nomenclature system for the tree of human Y-chromosomal binary haplogroups. Genome Research, 2002. doi:10.1101/gr.217602 245 markers genotyped; 153 binary haplogroups; a hierarchical nomenclature (abstract).
- van Oven M. PhyloTree.org: mtDNA tree Build 17. PhyloTree.org, 2016. Build 17 of 18 February 2016; 'currently comprising over 5,400 nodes (haplogroups)'. Read 10 October 2026.
- YTree v14.06.00 and its tree statistics. YFull, 2026. Release of 22 September 2026: 69,676 haplogroups and 459,477 SNPs in all (www.yfull.com/tree-build-stats/); haplogroup N 2,376 subclades; the M46 branch is labelled N-TAT. Read 10 October 2026.
- van Oven M, Kayser M. Updated comprehensive phylogenetic tree of global human mitochondrial DNA variation. Human Mutation, 2009. doi:10.1002/humu.20921
- Homo sapiens chromosome Y, GRCh38.p14 Primary Assembly (NC_000024.10). NCBI Nucleotide, 2026. Sequence length 57,227,415 bp, read from NCBI E-utilities esummary on 10 October 2026.
- Schönherr S, Weissensteiner H, Kronenberg F, Forer L. Haplogrep 3: an interactive haplogroup classification and analysis platform. Nucleic Acids Research, 2023. doi:10.1093/nar/gkad284 Kulczynski distance and a sorted list of hits; array input narrows the search range to the positions in the file; the most updated tree it cites holds 6,401 haplogroups.
- Kulczynski measure. Haplogrep 3 documentation, 2026. (HaplogroupWeight + SampleWeight) × 0.5, each the weight of the found polymorphisms over that of the expected or the sample polymorphisms. Read 10 October 2026.
- Weissensteiner H, Pacher D, Kloss-Brandstätter A, et al. HaploGrep 2: mitochondrial haplogroup classification in the era of high-throughput sequencing. Nucleic Acids Research, 2016. doi:10.1093/nar/gkw233
- Lamnidis TC, Majander K, Jeong C, et al. Ancient Fennoscandian genomes reveal origin and spread of Siberian ancestry in Europe. Nature Communications, 2018. doi:10.1038/s41467-018-07483-5 Uses N1c and N1c1a1a (N-L392); the West Eurasian cline 'spans from the Bedouins ... to north-eastern Europeans including Lithuanians, Norwegians and Finns'.
- Karafet TM, Mendez FL, Meilerman MB, Underhill PA, Zegura SL, Hammer MF. New binary polymorphisms reshape and increase resolution of the human Y chromosomal haplogroup tree. Genome Research, 2008. doi:10.1101/gr.7172008 311 distinct haplogroups, approximately 600 binary markers (abstract).
- Ilumäe AM, Reidla M, Chukhryaeva M, et al. Human Y chromosome haplogroup N: a non-trivial time-resolved phylogeography that cuts across language families. American Journal of Human Genetics, 2016. doi:10.1016/j.ajhg.2016.05.025
- Preussner A, Leinonen J, Riikonen J, Pirinen M, Tukiainen T. Y chromosome sequencing data suggest dual paths of haplogroup N1a1 into Finland. European Journal of Human Genetics, 2025. doi:10.1038/s41431-024-01707-7 'N1a1 (also known as N1c1, N3)'; branches compared with ISOGG v15.73 and YFull v10.01; YFull-style names; N1a1 in 64.3% of Finnish men.
- How many people have ever lived on Earth?. Population Reference Bureau, 2022. About 117 billion births in all; world population 450,000,000 in the year 1200 (Table 1). Read 10 October 2026.
- Poznik GD, Xue Y, Mendez FL, et al. Punctuated bursts in human male demography inferred from 1,244 worldwide Y-chromosome sequences. Nature Genetics, 2016. doi:10.1038/ng.3559
- Coverage: what each file type can tell you. Aimosti, 2026.
- Supported file formats: gVCF, VCF, chip, BAM, CRAM. Aimosti, 2026.
Last reviewed . Every number on this page links to the source it comes from; if one of them has moved, tell us.