The Ancestors of the Biobank: How 'Stairway Plot v3' Scales Up the Search for Ancient Indian Population Bottlenecks

Estimating the historical size of a population from modern DNA is fundamentally an exercise in reading the accumulation of genetic mutations over generations. When a population undergoes a sudden crash, or bottleneck, the diversity of its gene pool is drastically pruned. If a small surviving group then remains isolated—practically practicing endogamy (marrying only within a specific social group or community)—this restriction of the gene pool leaves highly distinct genomic signatures.

In South Asian demographic history, these genetic bottlenecks are particularly pronounced. Evidence shows that a significant proportion of the subcontinent's diverse groups have experienced severe founder events. These events are historical episodes where a small group of ancestors became genetically isolated, leading to high levels of shared chromosomal segments among their descendants today. Identifying the exact timing of these bottlenecks, however, has historically pushed genomic software to its limit.

A newly released software package, Stairway Plot v3, designed by researcher Xiaoming Liu at the University of South Florida, introduces a mathematical shift that allows geneticists to scan hundreds of thousands of sequences simultaneously on basic computer hardware. By making it computationally feasible to parse massive genomic biobanks, the tool opens up a more precise chronological window into when modern South Asian populations split, collapsed, or became endogamous.

The Bottleneck of Genomic Computation

The mathematical challenge in historical demography is finding the "effective population size," denoted as $N_e$. This term refers to the size of an idealized population that would lose genetic diversity at the same rate as the actual population under study. Fluctuations in $N_e$ over time trace a population's historical trajectory.

To reconstruct this trajectory, geneticists rely on the site frequency spectrum (SFS), which is a histogram showing the distribution of genetic mutations across different frequencies in a sample. Rare mutations—those found in only one or two individuals—usually represent very recent evolutionary events. Common mutations reflect deeper, older history.

Previous computational models have struggled with a major trade-off. Parametric models require researchers to pre-specify a historical scenario, such as estimating population size assuming a single bottleneck occurred exactly 2,000 years ago. While simple to compute, this method introduces human bias. Nonparametric models avoid this by letting the data speak for itself, but they have traditionally been incredibly slow. Older nonparametric algorithms, such as sequentially Markovian coalescent methods, are optimized to process only a few genomes at a time. Trying to run these programs on massive datasets of thousands of individuals would stall standard servers for days or weeks.

As Xiaoming Liu reports in the preprint, the breakthrough in Stairway Plot v3 lies in how the software reformulates this calculation. Instead of relying on slow differential-evolution optimizers, it recasts the demographic reconstruction as a mathematically convex optimization problem. It also breaks down large datasets by running analyses across multiple sub-sampled projections.

This mathematical redesign allows the program to scale up to 100,000 haploid sequences—which are single sets of chromosomes inherited from one parent, such as those found in sperm or egg cells, rather than the double sets found in most body cells—using standard CPU hardware without requiring expensive graphics processing units (GPUs). During testing on simulated data, Stairway Plot v3 processed 1,000 sequences in roughly eight minutes, whereas its predecessor, Stairway Plot 2, took nearly an hour, and older coalescent-based methods took tens of hours (Liu, 2026).

Mapping the South Asian Grid

This computational scale-up has direct implications for South Asian archaeogenetics (the study of ancient history using genetic materials). Genetic data from the subcontinent reveals a unique social tapestry. Rather than a single massive, homogeneous population, South Asia functions genetically as an amalgamation of thousands of distinct, endogamous groups.

Previous pioneering studies, such as those conducted by David Reich's lab and collaborators, demonstrated that many Indian communities underwent extreme founder events that were genetically more severe than those observed in Ashkenazi Jewish populations. For many of these groups, genetic evidence shows that the bottleneck occurred within the last 2,000 to 3,000 years.

⚖ Scholarly debate: While genetic data indicates that major bottlenecks occurred between 2,000 and 3,000 years ago, scholars debate the precise social and historical forces that drove this shift toward strict endogamy. One prominent school of thought suggests that this period coincides with the crystallization of the caste system and the formalization of social hierarchies in ancient texts, which legally and socially restricted intermarriage. Conversely, other scholars debate this timeline, arguing that ecological shifts, the collapse of Late Harappan urban centers, localized agricultural disruptions, or migrations of new populations played a more significant role in isolating communities before these social codes were widely enforced.

With older software, calculating the historical trajectories of hundreds of distinct ethnic, caste, and tribal groups required massive simplification. Researchers had to lump diverse populations together or rely on small sample sizes that missed the crucial rare genetic variants necessary to date recent history. Because Stairway Plot v3 can effortlessly scale to massive cohort sizes, geneticists can now use large-scale biobanks to analyze individual communities with extreme precision.

The tool's capabilities were highlighted in its preprint, where it was applied to a dataset of 54,302 Japanese individuals—the largest sample ever utilized for nonparametric, SFS-based demographic inference. The software successfully resolved very recent, subtle shifts in population size that had previously been invisible to less scalable programs.

Applied to South Asian cohorts, this resolution enables researchers to pin down the exact generations in which specific endogamous practices became strictly established. Rather than broad, sweeping claims about the antiquity of social structures, the software permits a highly localized timeline. Scholars can determine if a population bottleneck coincides with known historical disruptions, such as the collapse of major urban centers, shifts in agricultural patterns, or the localized spread of specific cultural practices.

Letting the Data Speak

A persistent challenge in Indian historical genetics is the tendency to over-interpret genetic data to fit preconceived historical or nationalist narratives. By removing the need to pre-specify a demographic model, Stairway Plot v3 acts as an objective mathematical filter.

Because the algorithm does not "know" the history of the subcontinent, it cannot bend the results to match any specific linguistic or archaeological theory. It simply processes the distribution of modern mutations to map past fluctuations in $N_e$.

By processing larger sample sizes, the software also minimizes the margin of error for recent history. When small sample sizes are analyzed, rare variants are often missed entirely, which makes the calculated demographic trajectory for the last 1,000 to 2,000 years highly unstable. Stairway Plot v3’s ability to handle tens of thousands of individuals means these rare variants are captured in abundance, providing stable, reliable bootstrap confidence intervals (a statistical method for estimating the uncertainty of a calculation) for recent centuries (Liu, 2026).

While the software provides a powerful lens into the past, geneticists still caution that $N_e$ is an abstract statistical measure rather than a literal head count of ancient people. External environmental pressures, localized epidemics, or shifts in marriage customs can all alter $N_e$ in ways that require careful archaeological and historical context to interpret. Nonetheless, by solving the bottleneck of computational scale, Stairway Plot v3 ensures that our calculations can finally keep pace with the massive volume of genomic data now emerging from the subcontinent.