How Many Genes Does a Human Have?

A human has roughly 20,000 genes that carry instructions for making proteins. The honest answer depends on what you count as a gene, though, and the total climbs well past 20,000 once other kinds are included. Either way, it is far fewer than scientists once expected, and no more than a tiny worm has.1iScience. The G-value paradox: Why similar genes can lead to very different brains

The Best Current Count of Protein-Making Genes

The big gene catalogues are built by different teams, and each publishes its own total:

The three main catalogues agree on 19,268 genes.6PubMed Central. The state of the human coding gene catalogues Another 2,603 genes appear in some catalogues and not others.7PubMed Central. The state of the human coding gene catalogues So the core of about 19,300 is solid, and the argument is over a borderline fringe.

Here “gene” means a protein-coding gene: a stretch of DNA holding the recipe for a protein. That has been the classic meaning since the early 1960s.8PubMed Central. The Evolving Definition of the Term “Gene”

A borderline gene is hard to classify because some DNA stretches look like they could make a protein, yet no one has ever seen the protein. Each catalogue team has its own rule for how much evidence is enough. A gene can be added or dropped between versions as evidence changes, which is why even one catalogue’s number shifts slightly over the years.

That also explains why three catalogues can each be right and still differ. GENCODE relies heavily on careful manual review, while others lean more on automated pipelines, so borderline calls go differently.

A textbook figure may come from an older catalogue release, so two reputable sources can differ without either being wrong. If you need a number for a citation, write down the catalogue name and release, not just the count.

For homework or conversation, “about 20,000 protein-coding genes” is accurate and matches every main catalogue. Quoting 19,000 to 20,500 shows the real range. Figures like 25,000 or 30,000 come from early-2000s estimates and are outdated. If precision matters, say “protein-coding,” because other gene types change the total.

One gene can also yield more than one product. Each protein-making gene can be read in several versions, about 5 per gene on average in one catalogue.9PubMed Central. CHESS 3: an improved, comprehensive catalog of human genes and transcripts based on large-scale expression data, phylogenetic analysis, and protein structure On paper, GENCODE’s roughly 19,400 genes therefore produce about 93,000 distinct protein sequences.10PubMed Central. CHESS 3: an improved, comprehensive catalog of human genes and transcripts based on large-scale expression data, phylogenetic analysis, and protein structure But when actual proteins are measured in cells, few genes show solid evidence of more than one protein form.11PubMed Central. Alternative splicing may not be the key to proteome complexity

Why Early Guesses Were So Much Higher

The first estimate, in 1964, put the human count at 6.7 million genes.12PubMed Central. Between a chicken and a grape: estimating the number of human genes It assumed all human DNA was made of genes laid end to end, with no gaps.13PubMed Central. Between a chicken and a grape: estimating the number of human genes Genes turned out to be broken up by long non-coding stretches, and much of the genome is repetitive sequence.14PubMed Central. Between a chicken and a grape: estimating the number of human genes Even so, a 1990 Human Genome Project report still estimated 100,000 genes.15PubMed Central. Between a chicken and a grape: estimating the number of human genes

Around 2000, estimates ran from 60,000 to 140,000.16New York Times. Scientists Cast Bets on Human Genes; a Winner Will Be Picked in 2003 Scientists opened a sweepstakes, and the average of 228 bets was 62,598 genes.17New York Times. Scientists Cast Bets on Human Genes; a Winner Will Be Picked in 2003 The lowest bet, 27,462, ended up closest to today’s figure. The count has fallen steadily ever since as tools improved.18PubMed Central. Between a chicken and a grape: estimating the number of human genes

What changed between 100,000 and about 20,000? Sequencing let researchers see real gene structure for the first time, instead of inferring it from totals. Computer programs could then scan the letters for the patterns that mark a working gene and count them directly. Comparing human DNA with that of mice and other animals helped too, because real genes tend to be preserved across species while random sequence drifts. Each improvement tended to remove candidate genes that turned out to be false alarms, which is why the count kept falling.

Genes That Don’t Make Proteins

Some genes make RNA, a cousin of DNA, that does its job without ever becoming protein. Catalogues list about 20,000 to 30,000 of one type, long RNA-only genes.19PubMed Central. Long non-coding RNAs: definitions, functions, challenges and recommendations They are often active only in certain cells at certain stages, so they are easy to miss, and some researchers think the true number could be far higher.20PubMed Central. Long non-coding RNAs: definitions, functions, challenges and recommendations

Why does an RNA-only gene still count as a gene? The older idea of a gene as a recipe for a protein is narrower than what many biologists now accept, which is any DNA stretch whose product does a job. RNA-only genes can act as regulators, helping switch other genes on or off, even though they never become protein. Because the definition is a matter of agreement, two scientists can look at the same data and count differently.

The total therefore changes with what you include. Protein-making genes alone give about 20,000. Adding RNA-only genes roughly doubles that or more. Including broken copies would raise it again, though most scientists don’t. That is why headlines sometimes cite 40,000 or higher.

A pseudogene is a defective copy of a gene that no longer works normally.21PubMed. Overcoming challenges and dogmas to understand the functions of pseudogenes Humans have an estimated 12,700 to 14,100 of them.22PubMed Central. The GENCODE pseudogene resource That makes broken gene copies nearly as common as working protein genes, yet they are not counted as genes.23PubMed. Overcoming challenges and dogmas to understand the functions of pseudogenes

A newer discovery is tiny genes. Thousands of very short stretches of DNA make tiny proteins that catalogues long ignored.24PubMed Central. Evolutionary origins and interactomes of human, young microproteins and small peptides translated from short open reading frames Many of these have real jobs in cells, including signalling to the immune system.25PubMed Central. Pervasive functional translation of non-canonical human open reading frames Some may be added to future counts. Will the number change again? Better tools may add tiny or rare genes, but the protein-gene count is expected to stay near 20,000.

The 37 Genes in Your Mitochondria

Mitochondria, the parts of cells that produce energy, carry their own small ring of DNA, called mitochondrial DNA. It is 16,569 letters long and holds 37 genes.26CD Genomics. Analyzing Human Mitochondrial Genomes Through High-Throughput Sequencing Data Of those, 13 make proteins, and the other 24 make RNA.27PubMed Central. Complete Mitochondrial Genome Sequence of the Human Neuroblastoma Cell Line 751-NA You inherit mitochondrial DNA only from your mother.

Catalogue counts focus on the main chromosomes, so these 37 are generally not folded into the 20,000. The 13 mitochondrial protein genes would be a tiny addition anyway. They matter because they power cell energy production, and faults in them cause some inherited diseases.

Mitochondria have their own DNA because they are thought to descend from free-living bacteria that were taken in by ancestral cells over a billion years ago. They kept a small set of their own genes while most of the original bacterial genes moved to the main chromosomes. That is why mitochondria copy and manage their DNA somewhat independently of the cell nucleus.

The 24 RNA genes mostly produce the machinery mitochondria use to build their own proteins. That includes transfer RNAs, which carry building blocks, and ribosomal RNAs, which form the assembly site. The mitochondrion thus has a self-contained system for making its 13 proteins on site.

Humans Versus Worms, Wheat and Other Species

Humans have roughly the same number of protein genes as a 1-millimetre nematode worm.28iScience. The G-value paradox: Why similar genes can lead to very different brains Mice come close to the human count too.29PubMed Central. The consensus coding sequence (CCDS) project: Identifying a common protein-coding gene set for the human and mouse genomes Wheat, meanwhile, has about 124,000 genes, around five times the human figure.30University of Nevada, Reno. You’re more complex than a worm, and researchers now believe they know why The puzzle is that gene number does not line up with how complex an organism seems.31iScience. The G-value paradox: Why similar genes can lead to very different brains

The worm in question is a standard laboratory animal with a few hundred cells in its nervous system. It feeds, mates, senses chemicals and learns simple behaviours using a gene set of similar size to ours.

Complex species tend to have more complex genomes, not more protein genes.32iScience. The G-value paradox: Why similar genes can lead to very different brains The control of when and where genes switch on matters as much as the number. Similar gene sets can build very different brains through that regulation, and RNA-only genes may help fine-tune human development.

Wheat’s huge count has a different explanation. Modern bread wheat came from crosses between different grass species, and the offspring kept the full gene sets of each parent. That left wheat with three related sub-genomes, so many genes exist in three similar versions. Plants often carry several full genome copies from ancient merges like this. A lot of wheat’s count is repetition, not three times as many distinct jobs.

Does Everyone Have the Same Number of Genes

The first truly complete human genome, finished in 2022, added nearly 200 million DNA letters.33PubMed Central. The complete sequence of a human genome It revealed about 2,000 predicted genes, 99 of them protein-making.34PubMed Central. The complete sequence of a human genome Most of the new ones were extra copies of genes already known.35PubMed Central. SEGMENTAL DUPLICATIONS AND THEIR VARIATION IN A COMPLETE HUMAN GENOME

People also differ from one another. Comparing many people’s genomes found 1,115 extra gene duplications not in the old reference.36PubMed. A draft human pangenome reference About 2,700 duplicated genes vary in how many copies people carry.37PubMed Central. Pangenome discovery and characterization of human protein-coding duplicated genes Some repeating genes, like those making ribosomes, sit in long clusters on five chromosomes.38PubMed. Human ribosomal RNA genes: orientation of the tandem array and conservation of the 5′ end

A gene is normally present in two copies, one from each parent. Some genes sit in a stretch of DNA that has been duplicated or deleted in certain people, so one person might have three copies where another has two. That can change how much of the gene’s product the body makes. Both people still have the gene, so two people can differ by dozens of gene copies while sharing the same set of gene types. Extra copies are usually counted as one gene family, so the 20,000 figure describes a reference person.

How 20,000 Genes Make Millions of Antibodies

Antibody genes come as many separate pieces instead of one finished recipe.39PubMed Central. Different classes of genomic inserts contribute to human antibody diversity Each developing immune cell picks one piece from each group and joins them.40PubMed Central. Different classes of genomic inserts contribute to human antibody diversity Because many options exist in every group, the number of possible combinations is far larger than the number of pieces. The joins are also imprecise, so variety multiplies further.41PubMed Central. Different classes of genomic inserts contribute to human antibody diversity

After meeting an invader, immune cells mutate these genes further to sharpen their grip.42PubMed Central. V(D)J recombination, somatic hypermutation and class switch recombination of immunoglobulins: mechanism and regulation A few hundred gene pieces can therefore yield antibodies against almost any invader.43PubMed Central. V(D)J recombination, somatic hypermutation and class switch recombination of immunoglobulins: mechanism and regulation

The cutting and joining happens only in developing immune cells. Your skin or liver cells keep the original, unshuffled pieces, and so does the DNA you pass to children. The shuffling doesn’t change a person’s standard gene count, only the genes inside particular immune cells.