Multi-sample merging

If you've tested with more than one service — say 23andMe and AncestryDNA — you can combine those files for the same person. Different arrays cover somewhat different markers, so merging gives you broader coverage and lets you cross-check where the two agree.

There are two distinct steps: grouping samples under an individual, and (optionally) merging them into one combined sample.

Group samples under an individual

An individual is a container that links the samples belonging to one person. Grouping does not change any data — each sample stays independent and keeps its own results.

1. Create an individual (give it a name).
2. Link your samples to it (each sample is annotated on its own first).

This alone is useful for keeping a person's samples together. To actually combine them, continue to merging.

Merge two samples

Merging creates a new, combined sample — the union of variants from both sources — with extra columns recording where each call came from and whether the two sources agreed.

From the individual's page, once exactly two source samples are linked, choose Merge samples. Already-merged samples linked to the same individual do not count as sources, so the button remains available after a previous two-source merge. A short wizard walks you through:

1. Strategy — how to handle sites where the two sources disagree:
   • Flag discordant calls (recommended) — keep neither call at a conflicting site; mark it as a no-call and record both original calls so nothing is silently guessed.
   • Prefer the first source (S1) — keep the first selected sample's call at conflicts.
   • Prefer the second source (S2) — keep the second selected sample's call at conflicts.
2. Preview — a dry run shows the concordance summary (below) and an estimated duration.
3. Confirm — name the merged sample and start it. Annotation runs automatically, after which the merged sample appears like any other.

The concordance summary

The preview and the Concordance Report break the merge down into buckets:

Bucket	Meaning
match	Both sources gave the same call (or both were no-calls)
filled_nocall	One source had no call; the other's call was used
discordant	Both called, but disagreed (handled per your chosen strategy)
unique (S1 / S2)	The site was present in only one source
collapsed_rsid	Same position in both sources but under different rsIDs (collapsed into one)

The Concordance Report (from the merged sample) shows these totals plus a paginated table of the discordant loci — the exact sites where your two files disagreed, with gene, consequence, and ClinVar context so you can see whether any disagreement matters.

Provenance on every merged variant

In a merged sample, each variant records:

• source — which file it came from (S1, S2, or both), and
• concordance — its bucket (match / filled_nocall / discordant / unique).

You can filter the Variant Explorer by these, e.g. to show only the sites where your two sources disagreed.

Good to know

• Two source samples at a time. Merging combines exactly two unmerged source samples. A merged sample can remain linked without blocking another run, but pair selection from three or more source uploads isn't supported yet.
• Same genome build only. Merging matches sites by coordinate, so both files must be on the same build. 23andMe and AncestryDNA are both GRCh37, so this is normally fine — just don't try to merge files from different builds.
• Both sources must be fully annotated and current before you can merge; you'll be asked to re-annotate a stale source first.
• A merged sample is a fresh, independent copy. It isn't a live view of its sources — re-annotating a source later won't change it. Deleting a source also deletes any merged samples built from it.
• Tags and watched variants don't carry over automatically; after a merge you'll be offered the chance to re-apply watches.