Reading a cell’s history: How RETrace2 and AVITI™ are rewriting lineage tracing

Researchers at Altos Labs have developed a new way to reconstruct the developmental history of cells and high accuracy homopolymer sequencing on AVITI helped make it possible.

Every cell carries a history

Every cell accumulates a record of its lineage with a unique combination of somatic mutations. These molecular marks can reveal which cells are related and which diverged long before they took on a specialized identity.

For decades, scientists have worked toward reading this record at scale. But the challenge has always been the same: how do you find enough high-quality molecular markers and reduce technical noise to reconstruct cellular lineage?

A new preprint from researchers at Altos Labs San Diego, the Medical University of Vienna, and Karolinska Institute demonstrates major progress toward that goal. Their method, RETrace2, doesn't just improve on existing lineage tracing approaches, it rethinks the fundamental strategy, and provides resolution previously considered out of reach.

The problem with existing lineage markers

Retrospective lineage tracing by inferring cell history from naturally occurring mutations is appealing because it requires no complex experimentation or genetic modification. But the quality of any lineage reconstruction depends on the quality of mutations and high-resolution detection of those mutations used as markers.

Each candidate marker type comes with its own trade-offs: mitochondrial DNA can exhibit genetic drift, methylation epimutations can be obscured by shifts in cell state, SNVs have low mutation rates requiring whole-genome profiling, making studies prohibit expensive.

The authors of this pre-preprint previously developed RETrace v1 as a proof of concept for using microsatellites and DNA methylation as a promising alternative for reconstructive lineage tracing because they are stable, permanent, and have higher mutation rates than SNVs.

RETrace2: A smarter marker, a smarter method

Here, they build on their initial technology to develop RETrace2. The authors hypothesized that version 1 was limited by the use of di- to hexa-nucleotide microsatellites which exhibit lower mutation rates. On the other hand, homopolymers, or single-base stretches, mutate faster than their multi-nucleotide counterparts.

To test this, the researchers applied two distinct probe sets to a microsatellite-unstable cell line (HCT116) with a fully known ground-truth lineage: one targeting homopolymers, the other targeting conventional di- to hexa-nucleotide repeats.

The results were clear:

100% vs 64% phylogenetic tree accuracy using homopolymer markers vs di-to-hexa microsatellites

The information advantage of homopolymers was measurable. Linear regression analysis showed that far fewer homopolymer markers could achieve the same phylogenetic tree accuracy and accuracy improved as you add more markers.

Engineering a better protocol from the ground up

Choosing the right marker was only the first step. RETrace2 required an entirely reimagined experimental workflow to capitalize on these advantages. The team made systematic improvements across every stage of library preparation, including increasing DNA yield, increasing marker coverage, improving probe hybridization, and reducing PCR amplification cycles. Together, the authors were able to see a 21-fold increase in markers captured per cell and a 98-fold increase in shared markers per cell pair.

Platform Spotlight · Why AVITI

The sequencing challenge that only AVITI could solve

Historically, homopolymers are among the most challenging sequences for any sequencing platform to read accurately. In a lineage tracing context, a single miscalled homopolymer length looks identical to a true somatic mutation. Get enough of those wrong, and the reconstructed phylogenetic tree drifts away from biological reality.

The researchers tested this by designing synthetic oligonucleotide benchmark containing poly-adenine homopolymers of varying lengths (10, 15, 20, 25, and 30 base pairs), sequenced the same library on both an Illumina NextSeq 2000 and the Element Biosciences AVITI, and compared Read 1 to Read 2 for each molecule as a platform-agnostic accuracy measure.

AVITI — 93.8%
(Sequencing accuracy maintained at 30 bp homopolymer)

Illumina NextSeq 2000 — 55.7%
(Accuracy at equivalent length)

The performance gap was clear. While Illumina's accuracy held up on short 1-15 bp homopolymers, it dropped to 55.7% at 30 bp. AVITI, powered by Avidite Base Chemistry™ (ABC™), maintained high accuracy across the full range of homopolymer length, retaining 93.8% accuracy at 30 bp, a 38-percentage-point advantage when sequencing longer homopolymers.

By adopting AVITI, the researchers combined an improved biochemical marker choice and removed a major source of systematic error that could have affected a lineage tree built on less accurate sequencing data.

As the paper's authors state directly: "The adoption of the AVITI platform in RETrace2 is therefore beneficial for increasing the accuracy for the longer homopolymer targets."

Validating RETrace2: from cell culture to the organism

In vitro ground-truth benchmarking

Beyond testing this method in microsatellite-unstable HCT116 cells, the team also tested RETrace2 in a microsatellite-stable mouse fibroblast line (3T3-L1), which better approximates normal, non-cancerous tissue.

RETrace v1 had failed entirely in this setting, yielding only baseline random accuracy (~33%). RETrace2, however, captured a median of 35,404 markers per cell, successfully reconstructed lineage with 70.1% accuracy.

In vivo multi-organ lineage reconstruction

Finally, the team applied RETrace2 to a living biological system: Msh2-deficient mice, which carry a knockout of a key mismatch repair gene and therefore accumulate microsatellite mutations at an accelerated rate. This model provided the high somatic mutation density needed for high-resolution lineage tracing while maintaining normal development.

Using the RETrace method, they profiled cells across seven tissue regions spanning three organs. After sequencing, the team reconstructed a high-resolution phylogenetic tree that spanned all three organs simultaneously.

The results revealed the polyclonal architecture of mammalian development with clarity. Statistical permutation tests confirmed the tree topology reflected genuine biology: cells from the same tissue clustered as clonal neighbors at significantly higher rates than chance. Among the nine clades identified, one-third showed strong tissue-specific enrichment, meaning those early progenitor cells had already committed to a particular organ lineage. The remaining two-thirds were multipotent reflecting progenitors that had not yet resolved their fate.

Simultaneous cell-type identification from the same cells

A key feature of RETrace2's dual-omic architecture is that lineage and cell identity information are captured from the same single cell in the same experiment. The methylation profiling arm of the workflow, adapted from single-cell reduced-representation bisulfite sequencing (scRRBS), generated a median of 10,770 CpG sites per cell.

Mapping the cells to a methylome reference atlas, the team was able to assign cell type identities. Ultimately, the result was a lineage tree that identifies each cell’s ancestral history (where in the tree it sits) and its identity (what kind of cell it became).

What this means for the future of developmental biology

RETrace2 is a significant step toward a long-standing goal: being able to reconstruct the complete cellular history of a mammal, from fertilized egg to adult tissue, without any genetic engineering. Several features make this vision more attainable than ever before, including, the ability of AVITI to highly accurately sequence through homopolymers.

Explore AVITI for your research

See how homopolymer accuracy can power your next breakthrough in genomics, lineage tracing, or single-cell biology.

Learn About AVITI

Read the original research: Cheng P-C, Kamenev D, Kameneva P, Fitzpatrick C, Adameyko I, Kharchenko PV, Zhang K. High-resolution retrospective single cell lineage tracing with mutable homopolymers. bioRxiv. 2026. doi: 10.64898/2026.03.10.709901