Several nearby genetic signals may not be independent clues

Draft • Research checked 4 October 2026 • AI-assisted DomDNA editorial content. No independent clinical review or publication is claimed. General education, not individual medical advice.

A table of genetic associations can contain several rows from the same region. Counting each row as a separate piece of biological evidence may exaggerate what the table establishes. Nearby variants can be correlated because their alleles tend to occur together in a population.

This relationship is called linkage disequilibrium. It is about patterns across genetic data, not a diagnosis and not proof that every nearby variant has the same function.

Why rows can travel together

The NIH LDlink documentation describes non-random co-occurrence of alleles and the measures used to examine it. Correlation can make one variant informative about another in a reference population.

The NHGRI linkage glossary provides the related physical idea: nearby sequences on a chromosome are less likely to be separated by recombination than distant ones. Physical closeness and statistical correlation are connected concepts, but they should not be treated as identical labels in every setting.

A hypothetical research table contains three associated markers near each other. If their alleles are highly correlated in the analysed population, the rows may partly reflect the same underlying signal. The table does not establish three independent mechanisms merely by listing three identifiers.

This also does not prove that only one functional variant exists in the region. The appropriate interpretation requires further analysis. The point is to avoid assuming independence from row count.

A small arithmetic example

Imagine a hypothetical reference set where whenever allele A appears at marker one, allele B appears at marker two. Seeing both in a person's file would not provide two fully independent observations of an association. Much of the information overlaps.

Now imagine another reference set where the co-occurrence is weaker. The same pair of markers can carry a different degree of overlapping information there. A correlation estimate should therefore stay attached to the reference population and dataset.

Do not turn this example into a rule to delete nearby rows from a personal report. An analysis may already account for correlation, or may use multiple variants for a specific reason. The reader's task is to ask what the method does, not to invent a replacement scoring algorithm.

A metric is not a probability of disease

LDlink reports measures including squared correlation and another measure called D prime. Their values describe relationships between alleles. They are not the probability that a variant is harmful or that a person will develop a condition.

A value near the top of a correlation scale should not be paraphrased as “very high health risk”. The name of the metric matters as much as its numerical value.

The NHGRI human-variation fact sheet provides broader context on the different kinds of variation people carry. A shared stretch of variation does not automatically identify which change, if any, affects a trait.

This article does not use a real association result to select a causal gene. Its examples explain how evidence can overlap before any clinical or functional conclusion is made.

Read the analysis rather than count the headlines

If several reports cite the same region, check whether they analysed distinct participants, distinct endpoints and independent evidence. Different headlines can repeat one dataset or correlated signals within it.

For a score or research summary, look for an explanation of how correlated variants were handled. A longer marker list is not automatically a stronger or more comprehensive result. Methods can legitimately select representative variants, model correlation or analyse a region in other ways.

If the explanation is absent, record that absence. Do not assume the author counted every row independently, but do not claim that correlation has been handled merely because the output looks polished.

What you can do with this

Choose one crowded region in a research table and save its method section, reference population and correlation explanation. This is a useful reading exercise; it is not a personal risk calculation.

Your next question can be specific: “Are these separate signals after accounting for correlation, or several markers tagging a shared pattern?” That invites an answer about the actual analysis.

Recognising overlapping information helps you judge the strength of a claim without dismissing useful genetic research or inflating a row count into certainty.

For general lifestyle learning, DomDNA’s educational quiz does not calculate linkage disequilibrium or genetic risk scores.