Is most of our DNA useless junk?
Written by AI from saved sources. A script checks every quote and every figure against the saved sources, and a second AI compares each claim with its source. How we work and corrections
Most human DNA does not code for proteins; only about 1%–2% does. How much of the rest is useless is not settled. Some noncoding DNA controls genes or protects chromosome ends.
Why people believe it
The argument deserves a fair hearing. The authors of the 2014 review “The Case for Junk DNA” defend the concept. They use “junk” for DNA with no functional role at the level of the organism. They trace the idea to differences in genome size. They also point to the decay of copied genes and theories about how populations evolve. They also describe media attention to claims that ENCODE had overturned junk DNA. These are their accounts of the debate. They are not a survey of why readers believe it.
“Almost none of it makes proteins”
True: MedlinePlus says about 1% of DNA codes for proteins. An NHS glossary gives 2%. These are approximations, not estimates of usefulness. Noncoding DNA includes switches that control when and where genes work. It also includes repeated sequences called telomeres, which protect chromosome ends during copying. These examples show that “noncoding” does not mean “useless”. They do not prove that every noncoding sequence has a job.
“So about 90% is useless”
A 2014 study compared mammal genomes. The authors estimated that evolution was filtering out insertions and deletions in 8.2% of human DNA. The authors inferred that much DNA showing biochemical activity in ENCODE could be deleted without affecting fitness. They reached this conclusion by comparing genomes, rather than by deleting DNA in an experiment.
The estimate extends a model to zero evolutionary difference between genomes. This assumes the model applies beyond the species differences studied. The authors say inferred turnover and the resulting constraint estimate could be biased upward. Their confidence estimates leave out these and other potential model errors. A 2023 Zoonomia study compared 240 mammal species. It reported unusually little change in at least about 10.7% of human DNA. The study compared this with repeated DNA evolving without selection. That is a minimum estimate for DNA showing this pattern. It does not prove that the rest is useless.
“Didn't ENCODE show it all has a purpose?”
ENCODE's 2012 study looked at 147 cell types. It detected biochemical activity across 80.4% of the genome in at least one of those cell types. Its definition allowed a reproducible biochemical signal to identify a functional element. That measures activity. It does not show that every active sequence benefits the organism.
A 2026 ENCODE4 paper reports regions that might control genes. These regions cover 21% of human DNA. Of these human regions, 97% had at least one test in at least one cell setting. Under the sampled conditions, 28% showed significant activity in at least one test. The authors say this likely underestimates the share of these regions that have a function. This is because most tests covered a limited number of cell types. They also favoured detecting enhancer activity. The 28% counts regions that might control genes, not DNA letters. It is not the percentage of DNA that is useful.
Thinking tool: define function before counting it
Ask what “works” means. The 2014 paper “Defining functional DNA elements in the human genome” distinguishes three approaches. These detect molecular activity, find evidence of evolutionary selection, or change DNA to see what happens. ENCODE's 80.4% and the 2014 comparative-genome estimate of 8.2% answer different questions. Even changing DNA can miss a role if another sequence makes up for it. It can also miss effects that appear only in rare cells or particular environmental conditions.
Use this elsewhere: before comparing percentages, check what each test counted and what it could miss.
Official positions, with dates
These are educational statements. The MedlinePlus page was updated 19 January 2021. It says we do not completely understand which noncoding regions control genes or have other functions. NHS England's glossary was updated 29 May 2019. It likewise says the function of the remaining 98% is not fully understood. NHGRI's page was updated 8 November 2012. It said it was unknown whether all ENCODE-detected active DNA contributed to cellular function. That statement concerns the 2012 findings.
What people also ask
Is 90% of our DNA junk?
That is not an established measured fraction. The evolutionary estimates concern selection or conservation, with limits. They do not directly test every sequence's usefulness.
Why is 98% of our DNA called junk DNA?
The NHS's 98% describes DNA that does not code for proteins. In “The Case for Junk DNA”, “junk” means DNA with no functional role at the level of the organism. These categories are different.
How much junk DNA do humans have?
No single settled figure follows. The tests ask different questions. Function has no universal definition.
Why is the term “junk DNA” no longer used?
The sources show debate over the term. The 2014 paper “The Case for Junk DNA” explicitly defends the concept. Discovering useful noncoding regions does not establish that all noncoding DNA is useful.
Check it yourself (5 minutes)
- Open MedlinePlus Genetics: What is noncoding DNA?
- Find “Noncoding DNA does not”. Read that sentence and the examples nearby.
- Does “noncoding” mean no useful job, or no protein-making instructions?
What you should have found
“Noncoding DNA does not provide instructions for making proteins.”
Before you forward this: does the box at the top say exactly what the sources below say?
Sources
- NLM / MedlinePlus Genetics, What is noncoding DNA? (2021).
- NHS England Genomics Education Programme, Non-coding DNA (2019).
- NHGRI, ENCODE: Deciphering Function in the Human Genome (2012).
- ENCODE Project Consortium, An integrated encyclopedia of DNA elements in the human genome (2012).
- 2014 analysis, Defining functional DNA elements in the human genome (2014).
- 2014 comparative-genome study, 8.2% of the Human Genome Is Constrained: Variation in Rates of Turnover across Functional Element Classes in the Human Lineage (2014).
- Zoonomia Consortium, Evolutionary constraint and innovation across hundreds of placental mammals (2023; public abstract only).
- ENCODE, An expanded registry of candidate cis-regulatory elements (2026).
- 2014 review, The Case for Junk DNA (2014).