How many pseudogenes in the human genome?

There are somewhat less than 25,000 genes in the human genome and there are probably about the same number of pseudogenes.

Pseudogenes are sequences that resemble real functional genes but they contain mutations that render them non-functional. They are very real examples of junk DNA.

There are four kinds of pseudogenes. Duplicated pseudogenes arise from a gene duplication event when one of the original copies mutates. Duplicated pseudogenes retain all of the features of the original gene, including introns and adjacent regulatory sequences. The inactivating mutation may occur in the gene itself—for example in the coding region of a protein coding gene—in which case the pseudogene may still be transcribed. Duplicated pseudogenes are usually found adjacent to their parent gene.

Processed pseudogenes arise when the normal transcript is copied by reverse transcriptase and the DNA copy is reintegrated into the genome. Processed pseudogenes don't have introns or regulatory sequences and they are not near their parent gene. Most processed pseudogenes come from transcripts that are expressed in the germ line.

Unitary pseudogenes come from single copies of genes that were no longer essential in the ancestral genome. The gene was mutated and the mutant allele was not detrimental so the gene became a pseudogene. These pseudogenes will have introns and regulatory sequences and they may still be transcribed. Unitary pseudogenes are located at the same locus as functional, and presumably still essential, genes in other species.

Polymorphic pseudogenes are like unitary pseudogenes except that the pseudogene allele has not become fixed in the genome so both the functional gene and the pseudogene are present in the population.

It's often quite difficult to identify pseudogenes, especially those derived from non-coding genes. Most of the computer programs designed to detect pseudogenes will rely on comparing known protein coding regions from an established database with sequences that resemble those coding regions but have various mutations that prevent production of a full-length functional protein. They don't look for pseudogenes related to noncoding genes.

It is quite easy to find pseudogenes in bacteria but much more difficult in complex eukaryotes because most of a protein coding gene consists of introns and the coding DNA in the exons is quite short. Early attempts with the human genome gave results ranging from more than 18,000 to a little over 11,000 depending on the program that was used.

A comparison of the pseudogenes detected by various groups revealed surprisingly little overlap so the actual number of pseudogenes could have been as low as 7,000 (Harrow et al., 2012).

The GENCODE annotation in 2012 listed about 11,000 pseudogenes derived from protein-coding genes (Harrow et al., 2012). The latest Ensembl/GENCODE annotation predicts 15,204 pseudogenes but it's not clear whether this includes non-coding genes [Human assembly gene annotation].

A recently published paper describes a new program that found a lot more pseudogenes.

Cabanac, S., Dunand, C. and Mathé, C. (2026) P-GRe: An efficient pipeline for pseudogenes annotation. Genomics 118:111216. [doi: 10.1016/j.ygeno.2026.111216]

Formerly considered as part of “junk DNA”, pseudogenes are nowadays known for their role in the post-transcriptional regulation of functional genes. Their identification also contributes to a better understanding of gene evolution, particularly in relation to adaptive responses and the evolution of multigene families. Despite this, there is, to our knowledge, no fully automatic pipeline allowing annotation of the pseudogenes on a whole genome. Here, we propose a new software named Pseudo-Gene Retriever (P-GRe). This is a completely automated pseudogene prediction tool requiring only a genome sequence, its corresponding GFF annotation file, and a protein sequences file. The aligner miniprot has been integrated in our pipeline, because of its high speed and sensitivity. With several filtering and post-analysis steps P-GRe outperforms existing software, while being more sensitive and bringing the new capacity of annotating unitary pseudogenes.

For now, I'll ignore the gratuitous comments about functional pseudogenes because there's no data in the paper on that topic—this is simply a paper describing yet another algorithm to detect pseudogenes related to protein coding genes.

Cabanac et al. detected 28,790 pseudogenes and that's far more than the number currently annotated in the latest reference genome. This could be due to the fact that the reference database of vertebrate proteins contained 19 million entries and these included transposon proteins and, presumably, virus and retrovirus sequences. I don't think that fragments of degenerate transposon genes should really be counted as pseudogenes.

The interesting part of this paper is that the authors characterize the various types of pseudogenes in Arabidopsis and humans. The surprising part is the number of unitary pseudogenes (~5200).

The classic example of a unitary pseudogene is the remnants of a gene for one of the enzymes in the pathway for synthesizing vitamin C [Human GULOP Pseudogene]. Our ancestral primates lost that gene, probably because they got sufficient vitamin C from eating fruit. It's difficult to imagine that there would have been 5000 other genes in our vertebrate ancestors that have been subsequently inactivated in the lineage leading to humans.

Maybe the authors are detecting virus and transposon genes?

There's a disturbing tendency in the pseudogene literature to attribute function to what is obviously junk DNA. This tendency is based, in part, on the discovery of a tiny number of examples of pseudogenes that have secondarily acquired a function. This is a classic example of a logical fallacy known as "cherry-picking" where scientists over-generalize from a few "exceptions that prove the rule." [Different kinds of pseudogenes - are they really pseudogenes?] [Are pseudogenes really pseudogenes?].

---

Harrow, J., Frankish, A., Gonzalez, J.M., Tapanari, E., Diekhans, M., Kokocinski, F., Aken, B.L., Barrell, D., Zadissa, A. and Searle, S. et al. (2012) GENCODE: the reference human genome annotation for The ENCODE Project. Genome Research 22:1760-1774. [doi: 10.1101/gr.135350.111]