eduKateSG · Genetics · The Gold Standard of Education
Read runs of homozygosity as evidence of shared ancestry, not as a single-number diagnosis
Choose the explanation you need, or follow all twelve chapters in order.
Full chapter index · Independent practice · Genetics subject guide
Runs of homozygosity, or ROH, are continuous genome segments in which the two chromosome copies carry the same called allele across many neighbouring markers. They can reveal autozygosity: both copies descend from the same ancestral chromosome segment. The useful question is not simply whether ROH exist, but what their lengths, locations and detection rules can support about ancestry.
This guide starts from recombination and inheritance, builds an auditable ROH summary, and then separates recent parental relatedness from older population history, technical artefact and local biological processes. The worked examples are fictional and teach the reasoning rather than diagnose a person or population.
Continue from The Gold Standard of Genetics within The Gold Standard of Education: learning that transfers into life and the How X Works learning library.
Full chapter index
See the mechanism · Chapters 1–2
Build the worked model · Chapters 3–5
Test what changes · Chapters 6–8
Check evidence and limits · Chapters 9–10
Practise and continue · Chapters 11–12
Homozygous means that the two called alleles at a locus match. Autozygous adds a genealogical claim: the matching copies are identical by descent from a common ancestral copy. An ROH is an observed stretch of homozygous markers that may be used as a proxy for autozygosity under stated data and calling rules.
These are not interchangeable labels. A common allele can be homozygous without recent shared ancestry, while a truly autozygous segment can be broken by a genotyping error or missed variant. Keep the observation, inferred state and historical interpretation on separate lines.
Define the individual or population, genome build, marker set, quality filters and whether sex chromosomes are included. State whether the task concerns recent consanguinity, background population structure, recessive-variant search or demographic history. The same ROH catalogue cannot answer every question equally well.
Evidence check: for Separate homozygosity, autozygosity and ROH, preserve the observation before the interpretation. Record sample, genome build, callable denominator and task. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
Meiosis breaks inherited chromosomes into segments. A shared ancestral segment tends to become shorter across generations because recombination has more opportunities to cut it. Very long autozygous tracts are therefore more compatible with a recent common ancestor than equally supported short tracts, while numerous short tracts can reflect older shared ancestry, population size history and background linkage disequilibrium.
This is probabilistic, not a family-tree clock. Recombination varies across the genome and between meioses; genetic length in centimorgans is more directly tied to recombination than physical length in megabases. Marker density, gaps and ROH settings also change apparent endpoints.
Do not convert one longest tract into an exact relationship label. Use the genome-wide distribution with pedigree, ancestry and uncertainty. The evidence may support “substantial recent shared ancestry” without identifying which named relatives connect the parents.
Evidence check: for Let recombination explain tract length, preserve the observation before the interpretation. Record homozygosity, autozygosity and ROH separately. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
CHAPTER 3 OF 12 · Build the worked model
3. Fix the ROH calling contract before comparison
Record software and version, sliding-window or model method, minimum physical or genetic length, minimum marker count, marker-density rule, allowed heterozygous calls, missingness, gap threshold and pruning. Different contracts produce different ROH totals even from identical genotypes.
Array data need enough informative markers; sequence data need depth, genotype-quality and variant filters. A centromere, assembly gap or low-complexity region can create a long low-information interval. Require the evidence to span the segment rather than merely connect two distant homozygous markers.
Make parameter choice task-specific and predeclared. Short-tract analyses are especially sensitive to linkage disequilibrium and marker ascertainment. Long-tract analyses still need error tolerance because a single false heterozygote can split one real segment. Report sensitivity across reasonable settings instead of choosing the most dramatic result.
Evidence check: for Fix the ROH calling contract before comparison, preserve the observation before the interpretation. Record software, markers, gaps and error tolerance. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
Let L_ROH be the summed length of qualifying autosomal ROH and L_auto the callable autosomal length. The genomic inbreeding coefficient F_ROH=L_ROH/L_auto. If a fictional genome has 180 megabases of qualifying ROH across 2,800 callable megabases, F_ROH=0.064, or 6.4 percent under that exact definition.
Partitioning matters. Suppose 120 megabases lie in tracts longer than 16 megabases and 60 megabases in 2–16 megabase tracts. The total is unchanged, yet the long-tract component carries stronger evidence of recent shared ancestry. Preserve both total and length classes.
Do not divide by a convenient whole-genome size when large regions were not callable. Do not compare an array-based denominator with a sequence-based denominator silently. Round only after calculation and retain segment coordinates so the total can be reconstructed.
| Quantity | Value | Calculation | Meaning |
|---|---|---|---|
| Callable autosomes | 2,800 Mb | Quality-filtered denominator | Not whole reference genome |
| All qualifying ROH | 180 Mb | Summed non-overlapping tracts | Pipeline-specific burden |
| FROH | 0.064 | 180 / 2,800 | Realised genomic proportion |
| ROH over 16 Mb | 120 Mb | Long-tract component | Stronger recent-ancestry signal |
Evidence check: for Calculate genome-wide ROH burden, preserve the observation before the interpretation. Record every segment total, class and FROH division. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
CHAPTER 5 OF 12 · Build the worked model
5. Interpret length classes as overlapping evidence
Length bins are practical summaries, not natural historical boxes. A study may label 1–2, 2–4, 4–8, 8–16 and over 16 megabases, while another uses genetic length. State the bins and do not transfer a threshold without checking marker density and population linkage patterns.
Longer classes are enriched for recent ancestry; shorter classes integrate deeper population history. But a local low-recombination region can remain physically long, and a recent segment can be shortened by recombination. Treat the age signal as a distribution.
Compare the individual with a relevant reference population processed through the same pipeline. A total that looks high relative to one cosmopolitan sample may be ordinary within a population with a different size and marriage history. Reference choice is part of the model, not neutral background.
Evidence check: for Interpret length classes as overlapping evidence, preserve the observation before the interpretation. Record length as probabilistic ancestry evidence. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
CHAPTER 6 OF 12 · Test what changes
6. Connect ROH with pedigrees without forcing agreement
A pedigree coefficient is the expected proportion autozygous under recorded relationships. F_ROH is a realised genomic proportion under a calling contract. Siblings with the same pedigree expectation can inherit different realised segments; incomplete pedigrees can miss shared ancestors.
Use disagreement diagnostically. High pedigree expectation but low F_ROH may reflect chance, pedigree error or calling choices. High F_ROH with a low recorded coefficient may reveal unrecorded relatedness, population background or denominator problems. Neither measure automatically invalidates the other.
Plot pedigree expectation against F_ROH with uncertainty and length components. When validating, avoid treating related individuals as independent. Family structure can make a pipeline appear more predictive than it is.
Evidence check: for Connect ROH with pedigrees without forcing agreement, preserve the observation before the interpretation. Record genetic versus physical map and recombination. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
CHAPTER 7 OF 12 · Test what changes
7. Distinguish individual ancestry from population demography
Across a sample, the distribution of ROH number and length can reflect effective population size, bottlenecks, isolation, migration and mating patterns. A population with many short-to-intermediate ROH may carry a historical contraction signal even if few individuals show extremely long tracts.
Sample design matters. Hospital cohorts, breed registries or volunteers may not represent the population. Recent migrants and admixed individuals can have different marker informativeness and linkage patterns. Stratify transparently rather than interpreting a mixed histogram as one history.
Demographic models should reproduce data features beyond total F_ROH. Compare length spectra, allele frequencies, identity-by-descent sharing and site-frequency information where appropriate. Several histories can generate similar ROH summaries, so report model competition rather than one inevitable story.
Evidence check: for Distinguish individual ancestry from population demography, preserve the observation before the interpretation. Record pedigree expectation versus realised genome. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
Regions with unusually frequent ROH across individuals are sometimes called ROH islands and may be investigated for selection. They can also arise from low recombination, structural variation, marker ascertainment, reference bias or assembly problems.
First ask whether the region is callable and informative in every group. Check recombination maps, segmental duplications, copy-number variation and marker density. Replicate in an independent dataset using a compatible but not identical technology.
Selection claims need more than recurrence. Combine ROH evidence with allele-frequency or haplotype tests, functional evidence and a plausible comparison design. A locally common haplotype is not automatically adaptive, and a candidate gene inside a broad segment is not the cause.
Evidence check: for Audit local ROH islands carefully, preserve the observation before the interpretation. Record demography, sample strata and reference choice. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
At segment level, inspect genotype quality, heterozygous calls, depth, Mendelian consistency and endpoints. At individual level, rerun reasonable thresholds and report how F_ROH and length classes change. At population level, bootstrap individuals or families and preserve sampling strata.
Technical replicates can reveal platform-specific splits and joins. Sequence a subset when an array result carries clinical or conservation consequence. Keep variant-calling uncertainty distinct from uncertainty in the ancestry interpretation.
For group comparisons, match callable length and pipeline. Adjusting for principal components may help control structure in an association analysis, but it does not make every ROH phenotype exchangeable. State the estimand: total burden, long-tract burden, local overlap or trait association.
| Layer | Primary check | What it constrains | Common failure |
|---|---|---|---|
| Segment | Depth, quality, endpoints | False joins and splits | Low-information gaps |
| Individual | Threshold sensitivity | Burden stability | One preferred setting |
| Family | Pedigree and Mendelian checks | Inheritance coherence | Non-independent validation |
| Population | Matched pipeline and bootstrap | Reference distribution | Unmatched callability |
Evidence check: for Validate segments and uncertainty, preserve the observation before the interpretation. Record segment, individual and population uncertainty. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
CHAPTER 10 OF 12 · Check evidence and limits
10. State limits, ethics and common misconceptions
ROH do not measure a person’s worth, identity or cultural membership. They do not justify stigma about communities or family practices. Genetic ancestry is not the same as genealogical, social or legal identity, and population averages must not be projected onto an individual.
An ROH total is not a count of harmful mutations. It may increase the chance that rare recessive alleles are homozygous, but variant effect, penetrance and environment remain separate questions. Nor does the absence of long ROH prove that the parents are unrelated in every genealogical sense.
Privacy matters because long segments and family comparisons can reveal unexpected relationships. Use consent, access control and a plan for incidental findings. For clinical interpretation, use qualified genetics professionals and validated standards rather than an educational calculation.
Evidence check: for State limits, ethics and common misconceptions, preserve the observation before the interpretation. Record identity, privacy and recessive-risk limits. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
CHAPTER 11 OF 12 · Practise and continue
11. Practise a changed-context interpretation
Individual P has F_ROH=0.064: 0.043 lies in tracts over 16 megabases and 0.021 in shorter qualifying tracts. Individual Q also has F_ROH=0.064, but nearly all lies in 2–8 megabase tracts. Do not report identical histories. P’s distribution is more compatible with recent shared parental ancestry; Q’s may reflect older population history, subject to the same pipeline and reference comparison.
Now a denser sequence call splits one of P’s longest array tracts with several high-quality heterozygous variants. Recalculate rather than defending the old label. If long-tract F_ROH falls materially while the total changes little, narrow the recent-ancestry claim and document the platform effect.
Evidence check: for Practise a changed-context interpretation, preserve the observation before the interpretation. Record the equal-total different-length changed case. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.
Archive consent scope, sample design, genome build, callable denominator, platform, software, every ROH parameter, segment coordinates, physical and genetic lengths, length classes, F_ROH calculations, reference population, pedigree information, sensitivity runs and interpretation limits. Frequently asked questions: Is every ROH autozygous? It is evidence, not proof. Do longer tracts imply more recent ancestry? Generally, with uncertainty. Is F_ROH a mutation count? No. Can two equal totals tell different histories? Yes.
Continue with inbreeding load and lethal-equivalent tests, effective population size and bottlenecks, and migration, selection and local adaptation. For foundations and empirical context, read the open studies on runs of homozygosity in European populations and ROH, population history and consanguinity. Their thresholds and samples differ, which is precisely why the calling contract must travel with the result.
Evidence check: for Build the evidence route and continue, preserve the observation before the interpretation. Record the narrowest ancestry statement supported. Then state the inference, one credible alternative, and the additional observation that would distinguish them. This compact audit keeps a fluent explanation tied to evidence that another reader can inspect.
Changed-context transfer: repeat this chapter’s decision after changing one consequential condition while holding the others fixed. Calculate or predict what should change, name the threshold or mechanism, and compare with the original case. If the conclusion survives, state the range tested; if it reverses, explain why. Transfer is demonstrated by the changed case, not asserted from the first example.
Practical decision record: write one sentence for what is known, one for what remains uncertain, and one for the next safe action. Include the evidence owner, the latest useful decision time, and the condition that would stop or reverse the action. This turns a technical result into a reviewable choice without pretending uncertainty has disappeared.