How much DNA do humans share with other animals?
Genome alignment, shared genes, and protein identity across 21 species
Matic Broz, PhD18 min
Pigs share 98% of our DNA, cats 90%, dogs 84% and bananas 60%, according to lists repeated across the web. The figures measure different things, from matching DNA letters to counts of related genes, and some have no traceable source at all, so the list cannot rank the animals. Measured the same way for every species, a typical pig protein is 89.1% identical to its human counterpart. Dogs, cats and cows give almost the same value, mice and rats slightly less, and chimpanzees 99.3%.[1]
Only the great apes come close to 98% at the DNA level. Chimpanzees and bonobos are our closest living relatives, and where the human and chimpanzee genomes line up, about 98.8% of DNA letters match.[2] Less than half of the human genome, 48.5%, lines up with the pig genome at all.[1] Table 1 gives two measurements for the species people ask about most, each column taken from one database for every row.
| Species | Protein identity | Shared genes |
|---|---|---|
| Chimpanzee | 99.3% | 91.2% |
| Dog | 89.3% | 82.0% |
| Pig | 89.1% | 85.6% |
| Cat | 89.0% | 83.5% |
| Cow | 88.7% | 85.4% |
| Mouse | 86.9% | 88.6% |
| Rat | 86.3% | 85.6% |
| Banana | – | 18.0% |
The two columns answer different questions. Protein identity says how alike a typical shared gene is, while shared genes says how many human genes have a recognizable counterpart at all, however much it has changed. Mice share slightly more genes with us than pigs or dogs do, yet their proteins have drifted further. For bananas only the gene count is meaningful: 18.0% of human genes have a banana counterpart, far below the popular 50%.[3]
The 98% pig figure has no source in the pig genome literature. The 2012 pig genome study analyzed about 9,000 one-to-one orthologs shared by six mammals to study gene evolution, not to measure overall identity.[4] Pigs resemble humans in organ size and physiology, which makes them useful in medicine and transplantation research, but anatomy cannot establish a DNA percentage.
Why published percentages disagree
A useful comparison must specify whether it measures DNA sequence identity, alignment coverage, protein identity, or shared genes. These quantities have different denominators and cannot be combined into a single ranking of animals.
| Measurement | What is counted | Published example |
|---|---|---|
| DNA sequence identity | Matching DNA letters among the positions compared | About 98.8% in directly aligned human–chimpanzee DNA, 2005 |
| Alignment coverage | The fraction of a reference genome included in an alignment | About 40% of the human genome aligned confidently with mouse, 2002 |
| Protein sequence identity | Matching amino acids in compared proteins | 75.3% median amino-acid identity among chicken–human one-to-one orthologs, 2004 |
| Gene counterpart presence | Genes with a detected evolutionary counterpart, regardless of exact sequence identity | About 70% of human protein-coding genes had a zebrafish ortholog, 2013 |
The chimpanzee figure comes from the 2005 draft-genome comparison, and the mouse alignment figure from the 2002 mouse genome study.[2][5] Sequence identity asks how often two lined-up letters match. Alignment coverage asks how much of a genome can be lined up in the first place, which becomes the limiting question for distant animals.
An ortholog is a gene in another species descended from the same ancestral gene through a species split. Orthologs may retain similar functions despite changes in their DNA and protein sequences. Gene duplications also mean that one human gene can correspond to several genes in another animal. Ensembl distinguishes these one-to-one, one-to-many, and many-to-many relationships when it reconstructs gene-family histories.[8][9]
The denominator matters even when both results count genes. The percentage of human genes with a zebrafish counterpart need not equal the percentage of zebrafish genes with a human counterpart. The species have different gene sets, and duplicated genes affect the two directions differently.
Which animals have the closest DNA to humans?
Chimpanzees and bonobos are humans' closest living relatives. The 2012 bonobo study reported 98.7% identity to humans in corresponding single-copy autosomal regions, meaning regions outside the sex chromosomes represented once in the comparison. The 98.8% chimpanzee figure and this 98.7% bonobo figure do not establish that chimpanzees are closer to humans than bonobos. The chimpanzee and bonobo lineages separated after their common ancestor had diverged from the human lineage. Differences in reference individuals, included regions, and analysis methods also affect the reported decimals.[10]
Monkeys are more distant relatives. In the 2007 rhesus macaque genome study, aligned human–macaque nucleotide sequences averaged 93.54% identity. Including small insertions and deletions reduced the reported identity to 90.76%. Changing the treatment of alignment gaps changes the answer even for the same two species.[11]
Ensembl's comparisons show the same broad pattern. In release 116, 91.4% of the human reference genome lies within the alignment with chimpanzee, compared with 90.3–90.4% for bonobo and gorilla, 89.1% for orangutan, and 82.0% for rhesus macaque. The typical one-to-one ortholog protein is 99.3% identical in chimpanzees, 98.9% in bonobos and gorillas, 98.4% in orangutans, and 97.0% in macaques.[1] The reference length includes about 151 million bases of unresolved gaps, roughly 5% of the human genome's 3.10-billion-base assembly, so even a perfect match could not reach 100%.[12]
More complete genome assemblies extend the comparison into regions that older drafts could not resolve. The 2025 complete ape genome study included repetitive and duplicated DNA and distinguished single-nucleotide differences from sequence that failed to align or did not fit a simple one-to-one alignment. Such regions cannot all be interpreted as ordinary mismatching DNA letters. The study broadens what can be compared; it does not turn the classic aligned-base percentage into a measurement of every base in both genomes.[13]
How similar are animal genomes to human DNA?
Published figures for different animals come from different decades, assemblies, and methods, so they cannot be ranked against each other. To compare animals on equal terms, we took all three measurements from a single source. Ensembl aligns genomes and infers orthologs for many species with the same pipelines, and release 116 covers every animal discussed here.[1][14] Across 21 species, gene sharing declines slowly with evolutionary distance, protein identity declines faster, and genome alignment collapses.
Genome alignment measures how much of the human reference falls inside Ensembl's pairwise alignment with each species. The alignment is built with LASTZ and then netted, so each human region keeps only its best-matching partner.[14] Among the non-primate placental mammals with an alignment, 32.5–55.5% of the human genome aligns. The figure falls to 4.6% for the egg-laying platypus, 2.5% for chicken, and 1.1% for zebrafish.
Coding DNA behaves differently. In the same alignments, 92.0% of human coding-exon bases align to mouse, 66.1% to chicken, and 53.4% to zebrafish.
Most non-coding DNA accumulates changes without harming the organism, so after enough time it can no longer be recognized. Protein-coding sequences are held in place by selection. The 2004 chicken study made the same observation: even for coding exons, only about 75% aligned between human and chicken.[6] Readers who want to compare two assemblies of their own can use MUMmer4, which anchors exact matches and reports aligned blocks, rearrangements, and coverage for a whole-genome alignment.
Protein identity asks a narrower question: among human genes with exactly one counterpart, how many amino acids match? Ensembl aligns one canonical protein per gene across each gene family, using M-Coffee or, for large families, MAFFT. We used its identity value for the human protein: identical amino acids as a percentage of the human protein's full length, so gaps and unaligned ends count as differences.[9] The median describes a typical gene, but the spread is wide. For mouse, the middle half of one-to-one orthologs ranges from 77.2% to 93.9% identity.
| Species | Genome aligned | Any ortholog | One-to-one ortholog | Protein identity |
|---|---|---|---|---|
| Chimpanzee | 91.4% | 91.2% | 84.2% | 99.3% |
| Bonobo | 90.4% | 87.2% | 85.2% | 98.9% |
| Gorilla | 90.3% | 89.5% | 85.7% | 98.9% |
| Orangutan | 89.1% | 89.8% | 86.1% | 98.4% |
| Rhesus macaque | 82.0% | 87.0% | 81.8% | 97.0% |
| Dog | – | 85.5% | 78.1% | 89.3% |
| Pig | 48.5% | 87.5% | 80.6% | 89.1% |
| Horse | 55.5% | 85.1% | 78.0% | 89.0% |
| Cat | – | 81.8% | 76.2% | 89.0% |
| Cow | – | 87.1% | 79.8% | 88.7% |
| Elephant | 44.1% | 83.0% | 75.9% | 87.0% |
| Mouse | 32.5% | 87.9% | 81.1% | 86.9% |
| Rat | – | 86.4% | 78.7% | 86.3% |
| Dolphin | 46.2% | 78.8% | 75.1% | 84.1% |
| Platypus | 4.6% | 75.1% | 68.7% | 74.2% |
| Chicken | 2.5% | 69.3% | 63.0% | 71.5% |
| Western clawed frog | – | 71.9% | 64.0% | 65.3% |
| Zebrafish | 1.1% | 69.8% | 48.2% | 58.3% |
| Fruit fly | – | 44.6% | 16.2% | 39.1% |
| Nematode worm | – | 40.0% | 13.7% | 33.9% |
| Baker's yeast | – | 18.9% | 6.7% | 29.6% |
We queried Ensembl's public databases on September 25, 2026. Human genes are those with the protein-coding biotype on the reference chromosomes, and each counts once per species however many orthologs it has.
The gene counts reproduce our earlier calculation from Ensembl BioMart for all 15 species it covered. The methods and limitations, species summary (CSV), gene-level ortholog table (CSV), paired comparisons (CSV), tree-of-life counts (CSV), OMA shared-gene counts (CSV), and analysis script record every query and calculation.
Each measure also carries artifacts of the data behind it. Chimpanzees and bonobos are equally related to humans, yet 91.2% of human genes have a detected chimpanzee ortholog and only 87.2% a bonobo one. Gorillas and orangutans fall in between, and macaques trail mice.
Assembly completeness, annotation, genuine gene gains and losses, and orthology inference all influence such counts, and the table cannot identify which factor explains a particular difference.[9] The ordering is therefore not an evolutionary family tree; building one requires phylogenetic analysis of aligned sequences, for example with IQ-TREE, which infers maximum-likelihood trees and branch support from a multiple sequence alignment.
The human denominator is specific to the annotation and filters. OMA counts 20,430 human genes against Ensembl's 20,107, and its shared-gene figures in Table 1 differ from Ensembl's by up to 3.5 percentage points for the same species. A denominator should not be replaced with a total from another release or database while retaining the same numerators. Our guide to the number of human genes explains why protein-coding gene totals change as annotations are revised.
Are humans closer to pigs, dogs, cats, or mice?
By protein identity, humans are about equally similar to dogs, pigs, horses, cats, and cows: the typical one-to-one ortholog in each is 88.7–89.3% identical to the human protein. Mice and rats are slightly lower, at 86.9% and 86.3%. Genome alignment tells the same story more starkly: 48.5% of the human genome aligns with pig and 55.5% with horse, but only 32.5% with mouse.
Medians summarize different sets of genes, so we also compared the species gene by gene. For each human gene with a one-to-one ortholog in both dog and mouse, we asked which of the two proteins is more identical to the human protein. Across 14,869 such genes, the dog protein was closer for 73.7% and the mouse protein for 23.2%, with the rest tied. Cat, horse, pig, and cow give nearly the same split.
This result is counterintuitive, because rodents are the closer relatives. Humans, mice, and rats all belong to one group of placental mammals, the Euarchontoglires, and the dog genome study used dogs as an outgroup to it. The same study estimated that DNA changes accumulate at different rates along each lineage, slowest in humans and fastest in mice.[15]
The mouse genome study likewise found about twice as many neutral substitutions on the mouse lineage as on the human lineage since the two species split.[5] Faster change over the same time erodes both genome alignment and protein identity, so rodent DNA now looks further from ours than dog or pig DNA does.
Gene presence gives a different order. Mouse and pig have detected orthologs for 87.9% and 87.5% of human protein-coding genes, compared with 85.5% for dog and 81.8% for cat. Differences of a few percentage points largely reflect how completely each genome is assembled and annotated, which is why a gene-presence ranking should not be read as a ranking of DNA similarity.
The pig shows how large that effect can be. A 2025 USDA study manually assembled pig genes and found assembly or annotation errors in 41 to 53% of the 6,135 genes it checked across three recent pig genome builds. With its corrected gene set, when a gene was missing from one of the pig, mouse and human genomes, pigs were five times more likely than mice to have the human gene.[16]
Between cats and dogs, the head-to-head comparison is close to a tie. Of 14,239 human genes with a one-to-one ortholog in both, the cat protein is more identical to the human protein for 47.6% and the dog protein for 44.9%, and 7.5% are tied. The cat's lower gene-presence figure therefore says more about its genome annotation than about how far its genes have drifted from ours. In OMA the order of the two reverses, with orthologs for 83.5% of human genes in cats and 82.0% in dogs.
The 90% cat figure has a narrower basis than the claim that 90% of all cat DNA matches human DNA. The 2007 cat genome study identified putative cat genes that included homologs for about 90% of the annotated human genes in its comparison. A homolog is a gene related through shared ancestry; detecting one does not require every DNA letter to match. The draft assembly, annotation, and homology criteria also differed from those in Ensembl, where we find 81.8%.[17]
For dogs, the 2005 reference-genome study compared genes, aligned sequence, and conserved chromosome organization across dog, human, and mouse. It does not support treating an unlabeled popular figure such as "84% shared DNA" as a universal whole-genome identity.[15] Our figures for dogs have explicit definitions: 85.5% of human protein-coding genes have a detected dog ortholog, and the typical one-to-one ortholog protein is 89.3% identical.
For mice, the classic genome study could confidently align about 40% of the human genome at the nucleotide level, and reported a median amino-acid identity of 78.5% across 12,845 one-to-one orthologs.[5] Our Ensembl values, 32.5% of the genome and 86.9% protein identity, use a newer assembly, a different gene set, and a different alignment method, and our data cannot separate those effects. Both analyses agree on the main point: finding a counterpart for most human genes is compatible with aligning only a minority of the genome.
What do the chicken, fish, and fruit-fly percentages mean?
The often-quoted 75% chicken figure concerns protein sequence identity, not 75% identity across complete human and chicken genomes. The 2004 chicken genome paper reported a median amino-acid identity of 75.3% among chicken–human one-to-one orthologs, compared with about 88% for rodent–human pairs.[6] Our Ensembl median for chicken is 71.5%. Only 2.5% of the human genome lies within the chicken alignment, yet 69.3% of human protein-coding genes have a chicken ortholog.
For zebrafish, the 2013 reference-genome paper reported that 71.4% of human genes had at least one zebrafish ortholog, and 47% a one-to-one ortholog. In a separate disease-gene comparison, 2,601 of 3,176 human genes with morbidity descriptions in the study's OMIM dataset had a zebrafish ortholog, reported as 82%. That is a defined subset of human genes, not 82% of the human genome.[7] Our Ensembl 116 values are close to the 2013 figures: 69.8% any ortholog and 48.2% one-to-one.
The large gap between those two zebrafish numbers has a biological cause. The ancestor of teleost fish duplicated its entire genome, so many human genes correspond to two zebrafish copies and cannot have a one-to-one partner.[7] Neither estimate can be generalized to every fish species, and the median protein identity of 58.3% applies only to the zebrafish genes that kept a single copy.
For fruit flies, 44.6% of human protein-coding genes have a detected ortholog, but only 16.2% have a one-to-one ortholog, and the typical one-to-one protein is 39.1% identical. A headline such as "60% shared DNA" cannot be substituted for these results without establishing its gene set, matching method, and denominator.
The same pattern extends further. The nematode worm has orthologs for 40.0% of human protein-coding genes, and baker's yeast, a single-celled fungus, for 18.9%. In all three, only about a third of the detected counterparts are one-to-one, because gene families have duplicated or been lost independently in each lineage since the split. Genes shared with yeast must already have existed in the common ancestor of animals and fungi. The proteome sizes by species guide compares how many proteins each of these organisms encodes.
Do humans also share genes with plants and bananas?
Yes. Humans and plants retain genes inherited from distant common ancestors, including genes involved in basic cellular processes. However, a shared-gene percentage for one plant cannot describe all plants or trees.
Ensembl also runs a separate pan-taxonomic comparison that builds gene trees for representative species of animals, plants, fungi, and protists.[18] Using its release paired with Ensembl 116 and the same 20,107 human genes, we found orthologs for 34.8–36.8% of human protein-coding genes in each of five plants: moss, thale cress, grape, rice, and tomato. Only about 7% have a one-to-one plant ortholog, reflecting the many gene and genome duplications in plant lineages.
Relatedness alone does not set this order. Fungi are closer relatives of animals than plants are, yet fewer human genes have an ortholog in baker's yeast (24.8%) than in any of the plants. Gene loss in a lineage removes counterparts as surely as divergence does. The pan-taxonomic values also run higher than the vertebrate comparison for shared species: 56.2% of human genes have a fruit-fly ortholog here, against 44.6% above. The two pipelines use different species sets and gene trees, so counts should be compared only within one of them.
Bananas are not in Ensembl's pan-taxonomic set. OMA, the database behind the shared-gene column in Table 1, assigns banana orthologs to 18.0% of human protein-coding genes. In a 2020 analysis, Natasha Glover compared human and banana genes using several orthology methods and found approximately 17–24%.[19] The roughly twofold gap between these banana figures and Ensembl's values for other flowering plants shows how much the orthology method alone can shift such a count.
Taken together, the three measures describe one process seen at different depths. Over tens of millions of years, most of the genome drifts beyond recognition, while many genes persist and their proteins change slowly enough to stay recognizable across animals, fungi, and plants. A claim about shared DNA is only reproducible when it names the species, the reference data, the unit of comparison, and the treatment of unmatched sequence.
For a small set of related genes or proteins, a sequence alignment can answer a more limited question. Tools such as MAFFT or Clustal Omega perform multiple sequence alignment of submitted homologous sequences, and UniProt download retrieves the protein sequences to compare. Any identity calculated from those alignments applies to those sequences and the stated gap-counting rules. A species-wide estimate requires a genome-wide analysis with its own coverage and identity definitions.
