How you get your DNA matches here
AutoTreeGen is not a DNA testing company, so uploading your raw DNA file does not produce a match list by itself. Your matches live at the company where you tested: Ancestry, MyHeritage, FamilyTreeDNA, 23andMe or GEDmatch. You bring that list here: link your kit to yourself in a tree, then import the list on the kit. The platform then estimates relationships and suggests the branch each match most likely belongs to; linking a match to a person in your tree stays your decision. Clusters need shared-match data: an AutoCluster export or a list with shared matches (ICW). The raw file still matters: it gives haplogroups and segments. Where a company has an export button (MyHeritage, FamilyTreeDNA, 23andMe, GEDmatch One-to-Many), download the CSV. Ancestry has no export: upload your raw file to GEDmatch (free) and take the One-to-Many CSV from there. We also show your AutoTreeGen Matches: other customers whose DNA overlaps yours.
Open:Get your DNA matches
What DNA is used for here
DNA is a separate evidence layer that sits beside the documents. It can support a documented line, contradict one, or point at a branch you have not worked on yet. It does not create relatives on its own, and it is never the only thing a conclusion rests on.
Open:Kits
Which DNA files we accept
The Add DNA picker asks what kind of file you have, not which company it came from. A raw genotype file from Ancestry, 23andMe, MyHeritage or FamilyTreeDNA - the provider is detected from the file's contents, never from its name. A GEDmatch kit number, which registers a kit by reference with no file. A match-list CSV from GEDmatch One-to-Many, MyHeritage, FamilyTreeDNA or 23andMe, which attaches to a specific kit rather than standing on its own. Ancestry does not offer a match-list export: upload your raw file to GEDmatch and take the One-to-Many CSV from there. An AutoCluster export from MyHeritage or Genetic Affairs. And Big Y results from FamilyTreeDNA, which are recorded on the person a kit is already linked to.
We do not fetch your DNA from anywhere
Entering a GEDmatch kit number records the reference - it does not make AutoTreeGen log in to GEDmatch and download anything. There is no connection to GEDmatch, MyHeritage, FamilyTreeDNA, Ancestry or 23andMe that pulls data for you. Everything the product analyses is a file you uploaded yourself.
Importing a match list
A match list belongs to one kit, so the import starts from the kit's row or from the Add DNA area with the file dropped first. Each kit button is labelled with its company, the person it is linked to, the kit number and whether a list was already loaded, so three kits with the same surname stay apart. The import window names the kit it is for, suggests the kit's company, shows a progress bar while thousands of rows are read and offers to open the matches when it is done. Any file you bring here is taken: the company is read from the file itself, segment and AutoCluster files go to their own pages for the same kit, and a file we cannot read yet is kept for review, never refused.
Open:Import DNA
Copy your matches from Ancestry or MyHeritage
Ancestry and MyHeritage do not offer a download of your DNA matches. You can copy them from the page instead. On the DNA kits page, open Import match list for your kit. Then: 1. Open your DNA matches page at Ancestry or MyHeritage. 2. At the bottom of the page, set the number per page to 50. On Ancestry it is called Items per page. On MyHeritage it is called Results per page. 3. Press Ctrl+A, then Ctrl+C. On a Mac, press Cmd+A, then Cmd+C. 4. Paste into the box Or paste from the page. 5. Go to the next page, copy it the same way and paste it under the last one. Do this for every page. Then tick the consent box and press Import. When it is done, you see how many matches were read and which company we found.
No CSV? Paste your match list from the company's page
You do not need a CSV file. Open the matches page at the company where the DNA was tested, scroll until the matches you want are loaded, press Ctrl+A (Cmd+A on a Mac) and Ctrl+C, then open Import match list for the kit and paste into the Paste your match list box. Saving the matches page as HTML and uploading that file works too. We recognise the company by content: Ancestry, MyHeritage, FamilyTreeDNA, 23andMe, GEDmatch One-to-Many and Geni. From the page we read each match's name, shared cM, number of segments, longest segment, predicted relationship and side when the page shows them; a saved HTML page also gives the profile link, so a second paste updates the same matches. Where a CSV exists it is still the fullest source: FamilyTreeDNA (Family Finder CSV export), GEDmatch (One-to-Many CSV). Ancestry and MyHeritage have no match download, so copying the page is the way. If the text is not recognised, the import says what kind of text arrived and what we need: a list with a name and shared cM (or shared DNA %) for each person. The paste follows the same size limit as a file, 150 MB.
Uploading a large match list
One match-list CSV can be up to 150 MB, so there is no need to split a big export. A list of 251,199 matches (96 MB) goes in with one upload. Open DNA, pick the kit and choose Import a match list. Pick the platform the file came from, choose the CSV, tick the consent box and press Import. The file is taken in seconds. Then it is processed in the background in six stages: Upload, Virus scan, Parse, Match, Write and Done. Live counters show matches read, matches prepared and the time left. You can close the window; the work goes on. A newer export of the same kit updates the matches already there, with no duplicates.
Reading a kit's matches
A kit's matches page shows each match with the full name, in compact rows. The pager sits above and below the list. Min cM is empty by default, so every match is shown; the count says how many pass the filter out of the kit's total. Where the export has it, a Match side filter and icons show which parent's side a match is on. Ancestry groups and the star appear under the name. The predicted relationship is a chip. The row menu opens the match here or on its platform.
Match groups and notes
On a kit's matches page you can sort matches into your own groups. Open Add / edit groups from a row menu: create a group and pick its color, rename a group (leave the name empty to bring back the name from your export) or delete a group you no longer need; the matches themselves stay. Tick matches, or select the whole page, to add many of them to a group or remove them at once. Each row shows its groups, and the group filter shows one group at a time. A note on a match, added from the row menu, stays when the list is imported again, and Review status lets you mark a match as reviewed or needing a closer look. Groups help you sort; they never change any analysis result.
A group label can stand for a branch
Many researchers label matches with letters or colours instead of surnames, for example G, B or S for each grandparent. The platform cannot read a branch from a letter, so in the group panel you can tie a label to an ancestor in your tree. From then on every match carrying that label is placed on that ancestor's branch automatically, with the label named as the reason. The link counts only when the ancestor is one of your own ancestors in the tree linked to the kit. Clearing the link removes the reason; nothing else changes.
Haplogroups and every column from your file
FamilyTreeDNA writes each match's own Y and mitochondrial haplogroups into the Family Finder list. They now appear as small chips on the match row; click one to see every match with the same haplogroup. The match card also has All columns from the file: the whole row exactly as your testing company wrote it, including ancestral surnames and notes. Nothing in that list is calculated by us.
Bring any file: we sort it for you
Upload doors take any file you bring. The file type and the company are read from the file itself, never from its name, so you do not need to know which box is the right one. Bring a DNA file, a segment file, an AutoCluster export or a document to the tree upload, and it is recognised and you are pointed to the section where it belongs. Bring a segment file or an AutoCluster export to a match-list import, and it is added to Segments or AutoClusters for the same kit, with a message saying so. A file we cannot read yet is kept for review rather than refused, so nothing you bring is lost.
A match is placed on a branch, and you decide
After every match-list import and once an hour the platform looks at each match of your kit and asks which branch of your tree it belongs to. A branch is a real ancestor in your tree, never a label we invent. The evidence is listed in the chip next to the match: a relative's kit that shares the same person, a parent's kit, a cluster edge, or a line label you wrote yourself, each with its weight. Press the small button next to the chip to open the evidence and confirm or reject the branch. Confirming writes the ancestor's surname as a group on that match and freezes the branch; rejecting means the platform never proposes that branch for that match again; you can reopen either. On the ancestor's own page you see how many of your matches point to that line, and the link opens them. Nothing in your tree changes from a decision: the branch is evidence, not a new relationship.
GEDmatch kit numbers are remembered, and you are told what to check
Write a GEDmatch kit number in a match note, or import a list whose Notes column carries one, and the platform remembers it: the number appears as a chip on the match row and on the GEDmatch kits page with every place it was seen. Register your own GEDmatch kit on the kits page, and the page then names the pairs to compare one-to-one: for every match whose link is in doubt (marked for review, a branch awaiting your decision, or more than one branch fitting) it shows their kit against yours. The platform never reads GEDmatch itself; you run the comparison there and record what you found in the note.
Open:GEDmatch kits
Linking a kit to a person
A kit is data until it is attached to a person in a tree; after that the analysis knows whose genome it is looking at and where that person sits. Only someone born recently enough to have tested can hold a kit, so people born long before DNA testing existed are not offered as candidates.
The same person often appears under different names on different platforms - or one account manages several kits.
Why linking asks for a ROOT person
Relationship reasoning is measured from the home person of the tree, so a tree with no ROOT has no origin to measure from. If you see "Set yourself as the ROOT on this tree first", open the person who is you, use Set as ROOT, and return to the kit. Check that you set it on the tree you are actually working in - a ROOT belongs to one tree, not to your account.
One person, several platforms
The same human can hold a GEDmatch kit, a MyHeritage kit, an FTDNA kit and an Ancestry kit at once, and one account often manages kits for several relatives. Several kits can be linked to the same person in your tree, which is the correct thing to do - they are one biological individual tested more than once, not four different people.
What merging two kits does to the matches
When you merge two kits of the same person, the matches, shared-match links and segments of the merged kit move to the surviving kit, so every engine reads them as that person's matches once - the branch engine, the kit bridges and the cousin search included. If the same match was already on the surviving kit, that row is kept and only filled in with what it lacked - a note, a group label, a haplogroup; the second copy is marked as a duplicate, never deleted. Unmerge moves everything back exactly as it was. Nothing is deleted, and the merged kit keeps its raw file, its consent and its person link.
Your matches do not need an AutoTreeGen account
A match that appears in a CSV you uploaded is analysed from that file. The person behind it has never heard of AutoTreeGen and does not need to. Keep four things apart: an external DNA match is a row in your export; a person is someone in your tree; a kit is a genome; and an account is someone who signs in here. Linking a match to a person in your tree is a research decision you make, not an invitation anyone receives.
Open:Matches
Clusters
Clusters come from an AutoCluster export you upload, or are computed from the match lists already attached to a kit. A cluster points at a branch. It does not measure how closely you are related to the people inside it.
A cluster is a group of DNA matches who also share DNA with one another - usually one branch of your family.
Open:Clusters
What the clusters page shows
Each cluster row shows the family surnames its members carry, the companies the matches come from, the strongest and the middle shared cM, how many members are already in your tree, and the branch of your tree the cluster most likely belongs to. If no branch can be named yet, it says so. Y and mt haplogroups of members show as small chips. Open a row to see every member; links open in a new tab. You can give a cluster your own name, and it stays when the clusters are worked out again. Sort the list or filter it by surname to find a family fast.
Triangulation, Big Y and AutoCluster files
Three more kinds of file go in through Import match list on the kit. A GEDmatch Triangulation report (CSV) adds one match for each kit number in it, and the shared segments for each side. A FamilyTreeDNA Big Y Matches CSV adds your Y matches, with their haplogroup, the step difference and the paternal ancestor each match named. Upload it again later and the same matches are updated, not doubled. An AutoCluster export from MyHeritage is best uploaded as the zip file you downloaded. If you already unzipped it, upload the report file from inside (the .html file), or pick several files at once; the owner, the kit and the date are read from the report itself, so a renamed file still works. If something is missing, the page says in plain words what to upload instead.
The predicted relationship for each match
The Predicted column shows how you are likely related to a match. When the testing company gave a prediction, that is what you see, and we never change it. When the file has only the shared cM, for example rows that came from a cluster file, the platform works out the likely relationship from the shared cM using the Shared cM Project table. When there is no shared cM at all, the cell says Not enough data. A prediction is a range of possible relationships, not a fact: the same amount of DNA can mean a second cousin or a great-aunt.
"In your tree" and "DNA on AutoTreeGen" on the match list
Many of your matches are already people in your tree, even before you link them. The match list looks for each match in your own trees, starting with the tree you have open, and shows who it probably is, with the birth year. It also reads nicknames and birth surnames written in brackets, as Geni writes them. Someone born before 1920 or who died before 1990 is never offered, because they could not have taken a DNA test. When the same person also has a DNA kit in your account, the row shows that too. You still decide: open the person and link the match when you agree.
AutoTreeGen Matches: other customers who share DNA with you
AutoTreeGen Matches, in the DNA menu, shows the other people on this platform who share DNA with you, whichever company each of you tested with. A cousin is found in three ways, and the page says which one: both raw DNA files were compared here, the same kit number appears in your match list, or the same full name appears in your match list. A second list shows kits that share cousins with you, with how many cousins you both match and the strongest one. Every new DNA upload is compared with everyone else's, so the page grows by itself. A kit whose owner switched DNA matches Off is not shown, and yours is not shown to others if you switch it Off.
Open:AutoTreeGen Matches
Kit bridges: one person seen by two kits
When two kits match the same person, for example your kit and your mother's, that person is a bridge between them. Kit bridges, in the DNA menu, lists each such person, which kits see them, with the shared cM, and the family lines you already marked on any of those kits. If one kit has a line and another does not, the page suggests carrying it over. Nothing moves by itself: open the bridge menu and choose Add to one kit, Add to all unmarked kits, Mark doubtful or Remove a line. A bridge needs at least two kits. Kits of other customers can appear in a bridge too, unless their owner switched DNA matches Off; you only ever change your own kits.
Open:Kit bridges
The haplotree and the story of a haplogroup
Haplotree, in the DNA menu, places the Y and mitochondrial haplogroups of the people in the tree you have open, and of your matches, on the scientific reference trees: YFull for Y and PhyloTree for mt. Each branch shows how many people sit on it or below it, and the page says how each one was placed. A name that is not in the reference tree is listed as not placed; nothing is guessed. Click any haplogroup, on the haplotree or on a chip in a match list, and its story opens: the path from the root to this branch with the markers that define each step, the branches below it, and your people and matches on this branch. In a match list each chip carries the colour of its main branch, a shade lighter or darker for each subbranch.
Open:Haplotree
The paternal line (Y) block on a kit
A kit with Y-DNA has a Paternal line (Y) block on its page. It shows the deepest Y branch recorded for the kit's person, how many Y matches and STR markers are on file, and the closest Y matches, sorted first by SNP difference and then by STR difference, each with the paternal ancestor the match named. Matches are also grouped by surname, so you see which families share your direct male line. If the kit has no Y-DNA yet, the block tells you what to upload: the Big Y Matches CSV from FamilyTreeDNA through the match list upload, and the Big Y files in Add DNA.
Shared segments
DNA segments shows the pieces of chromosome you share with your matches, read from the segment files you upload for a kit. Pick the kits to compare and the page draws the shared segments per chromosome. A triangulated segment, carried by three linked kits, points at a common ancestor more strongly than a plain shared-match group. Without segment files the page is empty, and it says that is why.
Open:Segments
Endogamy
In endogamous communities - Ashkenazi, Mennonite, Acadian, island and village populations - people are related through several lines at once. Shared centimorgans then overstate closeness, many small segments reflect several ancestral routes rather than one, shared matches turn up across branches that are genuinely separate, and the usual relationship calculators mislead. The product keeps the possible paths open instead of naming one.
Shared DNA suggests how you might be related - but in endogamous families it makes relatives look closer than they are.
Haplogroups
Where a kit carries Y-DNA or mitochondrial results, the haplogroup is recorded on the person that kit is linked to, together with the markers it was read from; the kit's haplogroup page shows the Y and mt predictions in two separate panels and its evidence page lists the markers behind each verdict. A prediction from an array chip is a candidate until a dedicated test confirms it. A haplogroup describes a deep paternal or maternal line; it is not a relationship to a particular ancestor and does not by itself confirm a documented link.
Patrilineal is not paternal, matrilineal is not maternal
Everyone on your father's half of the tree is paternal; everyone on your mother's half is maternal - that is half of your ancestors each. Y-DNA does not test that half. It follows one line only: father, his father, his father, until the fathers run out. That is the patrilineal line, the same path a surname usually takes. Mitochondrial DNA follows the matrilineal line: mother, her mother, her mother, until the mothers run out - both sons and daughters carry it, but only daughters pass it on. On AutoTreeGen the Y and mt chips, the painted lines in the tree explorer and the Community line matches mean exactly these two outer lines, never a whole side. A great-grandmother on your father's side is not on your patrilineal line, and her mother is not on your matrilineal line; to reach their lines you test a relative who descends from them along the right chain of fathers or mothers. When a tested relative's line reaches a person in your tree, the platform paints that line on the tree and names the kit it came from - a line you can check, not a colour we guessed.
Does the paper line agree with the DNA
For any ancestor the platform can check the direct male line (Y-DNA) and the direct female line (mtDNA). It looks for your kits whose testers sit on that line in the tree and compares their haplogroups. Two or more kits with compatible haplogroups mean the line is consistent. Two confirmed kits that disagree mean the paper line is broken somewhere between them, and both kits are named. One kit, chip predictions only, or no kit on the line give no verdict, and the reason is said. A haplogroup copied down the tree from another person's kit never counts as a second witness.
The paternal line check
Whenever a Y-DNA result is recorded on a kit, the platform compares its haplogroup with the documented paternal line of the person the kit is linked to. A clash between a father and a son is written as a parentage assessment with the evidence on both sides and listed under Parentage checks; the import dialog tells you whether the check ran, found nothing, or found conflicts. It needs the consent recorded for the kit and a person linked to it. It never rewrites a parent.
Open:Parentage checks
Lineage ribbons and numbered lines
The Lineage page draws the paternal Y-DNA and the maternal mtDNA line of the person you are on this tree, ancestor by ancestor, with the haplogroup that a tested relative's kit gives each line and a status glyph on every step. Lines get a stable number: Y-1, mt-1 and so on, and the number stays with the line when its haplogroup is refined. A line nobody on it has tested is drawn empty. The tree's DNA page lists the relationship verdicts the DNA layer has reached for this tree, with the same honest states.
DNA and AI are separate
No language model invents a biological relationship from your DNA. The DNA layer is deterministic analysis of the files you uploaded. A language model may help you read and organise documented research context, and where it does, it is told so on that screen - but it never turns a shared segment into a cousin.
Privacy of your DNA
Genetic data is a special category under GDPR Article 9 and is treated that way: it is encrypted at rest, every upload records the consent it was given under - your own kit, or one you are authorised to manage - and that consent can be withdrawn. Your tree and your DNA are never sold, never shared with advertisers, and never used to train AI models. Your consents page lists what you have agreed to and lets you change it.
Open:Your consents
Step-by-step guides
Need help with this step?Ask the assistant