Start here

The big picture

Everything here serves one of two goals:

1. Build the tree — people, relationships, dates and places. Every tested kit gets attached to its person, so the tree and the DNA point at each other.
2. Fill in the ancestral DNA — when several matches all share the same piece of a chromosome, that piece can be accumulated onto the ancestor who passed it down. Those inferred pieces are called Tree DNA, and the Chromosome Map draws them per chromosome, coloured by ancestor.

The two goals feed each other in one loop:

↻ then repeat — every new relative sharpens the map

You will still use the other guides for the detail; this page is the map. If a term below is unfamiliar, the jargon buster at the bottom has one-line definitions.

1Get your DNA in

Goal: your raw DNA — and as many relatives’ kits as you can gather — uploaded and being compared.

1. Create a free account and confirm your email.
2. Download the raw data file from the company that tested you (see Downloading your file).
3. Upload it exactly as downloaded — .txt, .csv, .zip or .gz. Your kit gets a kit number and is compared against every other kit.
4. Repeat for any relative who has tested. Parents, siblings, aunts, uncles and cousins are each worth far more than another ethnicity estimate.

Why relatives matter: everything later — triangulation, phasing, reconstruction — gets easier with more tested people. Three matches on one segment is what turns a guess into evidence; a tested parent or sibling is what lets you separate your two sides.

2Meet your matches

Goal: a shortlist of your closest and most promising matches, and a feel for how cM and segments behave.

1. Open One-to-Many and read down the list: shared cM, segments and largest segment. The top of the list is your close family.
2. Click Review on a match for the full match page — relationship estimate, shared matches and a chromosome browser.
3. Click Compare (or open One-to-One) to see the exact segments you share.
4. Start labelling as you go: Tag Groups are your own private colour labels, and each match card takes a private note.

Reading the result: one big segment is stronger evidence than many tiny ones; segments under about 7 cM usually are not evidence of a real relationship. The matching guide explains cM, IBD1/IBD2 and the relationship ranges in depth.

3Check for endogamy

Goal: know how much to trust the cM numbers before you build on them.

Run Are Your Parents Related? once, early. It scans your kit for long runs of homozygosity (ROH) — stretches where both copies of a chromosome carry the same DNA.

• No significant ROH → the standard relationship ranges apply fairly directly.
• Long ROH → your parents share an ancestor, and so probably do many of your matches. Distant relatives will share more cM, in more segments, than the standard tables assume.

In Roma and other endogamous communities this is common and expected — it is community history, not a scandal. The practical takeaway: lean on the largest segment, triangulation and the tree rather than the cM total alone. See the endogamy notes in the matching guide.

4Split matches into family lines

Goal: turn a flat match list into groups that look like ancestral lines — then prove a group with segments.

1. The Leeds method groups your matches into (up to) four grandparent groups — the classic first way to split a match list into family lines.
2. Identify one known relative per group and the rest of the group is probably on that person’s branch. Label it with a Tag Group so the match list stays readable.
3. Segment Search lists every shared segment, so you can find everyone who shares a particular region.
4. Triangulation takes it further: it finds clusters of matches who all share the same segment with each other. That is the point where a segment becomes evidence of one common ancestor — and the tool can add it to your tree (step 7).

Hands off to: named or hypothesised lines, and triangulated segments ready to be painted onto ancestors.

5Build the tree and attach kits

Goal: a tree that contains both the people and the tested DNA, so every later tool has somewhere to put its result.

1. Open the tree viewer. Every registered user can edit the shared tree; right-click a card for the menu (edit, add relative, set home). It is modelled on the way Ancestry’s family tree works, though nowhere near as sophisticated — it should do the job for now.
2. Add the people you know — parents, grandparents, siblings, the aunts and uncles — and attach each tested kit to its person (from DNA Kits → kit settings → tree linking, or from the person’s page).
3. Set your own home person with the star, so the tree opens where you want.
4. DNA badges are on by default — every tested person who matches the tree’s selected DNA test carries a green DNA {cM} badge, and inferred Tree DNA people carry a blue one. The bottom rail lists that test’s top DNA matches: click a name to jump to them in the tree, or a dashed one to open its Review page. Then open Compare DNA to tree to see which branch a kit’s matches fit.
5. If the same person appears twice, use the Review & merge banner on the person page.

Hands off to: a tree with tested people wired in — which is what makes step 7 possible. The tree guide covers badges, ThruLines and merging in detail.

6Place the matches that do not fit

Goal: find where a mystery match belongs instead of guessing.

1. WATO (“What Are The Odds?”) scores every plausible place a match could sit in your tree. It compares the observed shared cM against what each placement would predict, using your real tested relatives as references, and ranks the placements.
2. Read the top placements, then add the person to the tree under the best-supported one (or leave them floating until more evidence arrives). A “No possible combination found” message means no placement explains the cM — do not trust the top row.
3. No tree at all yet? AutoKinship builds and scores possible arrangements from your matches alone.

Keep in mind: endogamy biases these estimates towards “too close”, and a predicted placement is a hypothesis until records or new testers confirm it. Neither tool is a self-learning model, and they do not both use the tree the same way: WATO is only as good as the tree — it scores against the tested relatives you have placed there, so every relative you add or placement you correct sharpens its next result — while AutoKinship works from your match list alone and does not depend on the tree at all. A thin or wrong tree gives WATO thin or wrong placements.

7Fill in your ancestors’ DNA the payoff

Goal: give the people in your tree the DNA they must have carried — even the ones who never tested.

The engine does this from evidence you have already collected. When a group of matches triangulates on one segment, the shared (ancestral) allele can be worked out, and it is accumulated onto the Tree DNA virtual kit of the ancestors the segment passed through. People who have a real kit never get inferred DNA — their real file is authoritative.

Pick the path that fits what you have:

Then look at what you built:

• Tree DNA reconstructions — every virtual kit, with its SNPs, segments and coverage.
• The ancestor’s person page — the inferred segments assigned to them.
• Chromosome Map — your kit’s chromosomes drawn with each ancestor’s segments in their own colour (X follows inheritance).
• Compare DNA to tree DNA — run a real kit against the reconstructions with the matching engine.
• Tree conflicts — segments that could not be added because they disagree with what is already there. If the same block keeps being deleted by one person and added back by another, it is marked contested and can be escalated to an admin (from Visual Phasing): both versions cannot be right, so one of the branches is incorrect and needs checking against records. Escalated blocks and the full paint/remove history are listed on the conflicts page.

8Keep the loop going

Goal: every new tester makes the map sharper — so keep feeding it.

• Invite relatives to test or upload. A new cousin can confirm a cluster, split a line, or extend a segment one generation further back.
• Re-run the grouping tools after new kits arrive: matching happens automatically, and clustering/triangulation recompute when you open them.
• Keep the tree and the DNA in sync — when a match is confirmed, attach them to their person; when a segment conflicts, review it.
• Side quests: Y-DNA and mtDNA follow your deep paternal and maternal lines; the Romani report and Ethnicity explore where your ancestry comes from. Those answer “where from” — this roadmap answers “who”.

Jargon buster

Raw DNA

The plain text file of your genotypes from the testing company — the file you upload, not the matches or ethnicity report built from it.

SNP

One position in the genome that a testing chip reads. Modern kits cover hundreds of thousands of them.

cM (centimorgan)

A unit of genetic distance shared with a match. More cM generally means a closer relationship — but endogamy inflates it.

Segment

One unbroken run of shared DNA, given as chromosome, start position and stop position.

IBD1 / IBD2

Shared on one chromosome copy (half-identical) / shared on both copies. Parent and child share everywhere at IBD1 with no IBD2; full siblings show both.

Triangulation

Three or more matches who all share the same segment with each other — strong evidence the segment came from one common ancestor.

Phasing

Working out which allele came from which parent, so a genotype can be split into a maternal and a paternal copy.

Tree DNA / virtual kit

A kit reconstructed for someone who never tested, built from descendants’ DNA. A hypothesis, not a test result; shown with a VK kit number.

Endogamy

Generations of marriage within one community, so distant relatives share more DNA than the standard relationship tables assume. Common in Roma families.

ROH

Runs of homozygosity — long stretches where both copies of a chromosome are identical, a sign that your parents share an ancestor.

Kit number

The short code for a kit: AN, MH, LI, FT, M3 for real uploads, VK for Tree DNA, SK for a superkit, LAZ for a Lazarus reconstruction.