The first printout of the human reference genome presented as a series of books, displayed at theWellcome Collection, London
Areference genome is agenome assembly that represents thecomplete genetic sequence of an organism as a continuous string ofnucleotides (A, T, C, and G). For an assembly to serve as a reference genome, it is typically accompanied by annotations, produced through a process known as DNA orgenome annotation. The annotations specify the genomic coordinates (start and end locations) ofgenes,exons,introns, andmRNA, and are often paired with corresponding transcript (mRNA) andprotein sequences (algorithm predicted or experimentally validated).[1]
Reference genomes exist for a wide variety ofspecies, including species ofviruses,bacteria,fungi,plants andanimals, and they differ in how they are constructed and represented. A reference may be derived from a single individual or from multiple individuals whose sequences are collapsed into one representative assembly -haplotype. Two main factors determine reference genome's assembly quality: thesequencing technology which affects sequence accuracy and the assembly level which indicates how complete the genome representation is.[2][3]
The ideal is a chromosome-level assembly, which is a complete DNA sequence for each chromosome with no unplaced segments. However, achieving this remains technically challenging, especially for large or repetitive genomes (dense inrepetitive elements). Earlier sequencing technologies often produced assemblies at thecontig (short contiguous sequences) orscaffold (ordered sets of contigs) level, with limited chromosomal context. The exact size of these fragments depends on the sequencing platform and bioinformatic methods available at the time.[4]
For assemblies that are not fully resolved, summary statistics such as N50 and L50 are commonly used to characterise contiguity and assembly fragmentation; these metrics are explained in theContigs andScaffolds section.
Reference genomes are central toomics research, particularlygenomics. They provide a reference for "mapping" DNA sequence data from many individuals, enabling efficient identification of the genomic location of these sequences and the detection ofpolymorphisms (sequence differences among individuals) through a process known asvariant calling.[5]
The limitations of this practice, such as reference bias and under-representation of population diversity, have led to the development of population-level reference sets andpangenomes.[6]
Reference genomes and their annotations are publicly accessible through online genome browsers and archives such asEnsembl,[7] the European Nucleotide Archive (ENA) atEMBL-EBI, theUCSC Genome Browser, andNCBI.
The length of a genome can be measured in multiple different ways.
A simple way to measure genome length is to count the number of base pairs in the assembly.[8]
Thegolden path is an alternative measure of length that omits redundant regions such ashaplotypes andpseudo autosomal regions.[9][10] It is usually constructed by layering sequencing information over a physical map to combine scaffold information. It is a 'best estimate' of what thegenome will look like and typically includes gaps, making it longer than the typical base pair assembly.[11]
Diagram of reads arrangement, formingcontigs and these can be assembled intoscaffolds in the complete process of sequencing and assembly of a reference genome. The gap between contig 1 and 2 is indicated as sequenced, forming a scaffold, while the other gap is not sequenced and separates scaffold 1 and 2.
Reference genomes assembly requires reads overlapping, creatingcontigs, which are contiguous DNA regions ofconsensus sequences.[12] If there are gaps between contigs, these can be filled byscaffolding, either by contigs amplification with PCR and sequencing or byBacterial Artificial Chromosome (BAC) cloning.[13][12] Filling these gaps is not always possible, in this case multiple scaffolds are created in a reference assembly.[14] Scaffolds are classified in 3 types: 1) Placed, whose chromosome, genomic coordinates and orientations are known; 2) Unlocalised, when only the chromosome is known but not the coordinates or orientation; 3) Unplaced, whose chromosome is not known.[15]
The number ofcontigs andscaffolds, as well as their average lengths are relevant parameters, among many others, for a reference genome assembly quality assessment since they provide information about the continuity of the final mapping from the original genome. The smaller the number of scaffolds per chromosome, until a single scaffold occupies an entire chromosome, the greater the continuity of the genome assembly.[16][17][18] Other related parameters areN50 andL50. N50 is the length of the contigs/scaffolds in which the 50% of the assembly is found in fragments of this length or greater, while L50 is the number of contigs/scaffolds whose length is N50. The higher the value of N50, the lower the value of L50, and vice versa, indicating high continuity in the assembly.[19][20][21]
The original human reference genome was derived from thirteen anonymous volunteers fromBuffalo, New York. Donors were recruited by advertisement inThe Buffalo News, on Sunday, March 23, 1997. The first ten male and ten female volunteers were invited to make an appointment with the project'sgenetic counselors and donate blood from which DNA was extracted. As a result of how the DNA samples were processed, about 80 percent of the reference genome came from eight people and one male, designatedRP11, accounts for 66 percent of the total. TheABO blood group system differs among humans, but the human reference genome contains only anO allele, although the others areannotated.[22][23][24][25][26]
Evolution of the cost of sequencing a human genome from 2001 to 2021
As the cost ofDNA sequencing falls, and newfull genome sequencing technologies emerge, more genome sequences continue to be generated. In several cases people such asJames D. Watson had their genome assembled usingmassive parallel DNA sequencing.[27][28] Comparison between the reference (assembly NCBI36/hg18) and Watson's genome revealed 3.3 millionsingle nucleotide polymorphism differences, while about 1.4 percent of his DNA could not be matched to the reference genome at all.[26][27] For regions where there is known to be large-scale variation, sets of alternateloci are assembled alongside the reference locus.
Chromosomes ideogram of the human reference genome assembly GRCh38/hg38. Characteristic bands patterns are displayed in black, grey and white, while the gaps and partially assembled regions are displayed in blue and rose, respectively. Reference: Genome Data Viewer of the NCBI.[29]
The latest human reference genome assembly, released by theGenome Reference Consortium, was GRCh38 in 2017.[30] Several patches were added to update it, the latest patch being GRCh38.p14, published on the 3rd of February 2022.[31][32] This build only has 349 gaps across the entire assembly, which implies a great improvement in comparison with the first version, which had roughly 150,000 gaps.[23] The gaps are mostly in areas such astelomeres,centromeres, and longrepetitive sequences, with the biggest gap along the long arm of the Y chromosome, a region of ~30 Mb in length (~52% of the Y chromosome's length).[33] The number ofgenomic clone libraries contributing to the reference has increased steadily to >60 over the years, although individualRP11 still accounts for 70% of the reference genome.[34] Genomic analysis of this anonymous male suggests that he is of African-European ancestry.[34] According to the GRC website, their next assembly release for the human genome (version GRCh39) is currently "indefinitely postponed".[35]
In 2022, the Telomere-to-Telomere (T2T) Consortium,[36] an open, community-based effort, published the first completely assembled reference genome (version T2T-CHM13), without any gaps in the assembly. It did not contain a Y-chromosome until version 2.0.[37][38] This assembly allows for the examination of centromeric and pericentromeric sequence evolution. The consortium employed rigorous methods to assemble, clean, and validate complex repeat regions which are particularly difficult to sequence.[39] It used ultra-long–read (>100 kb) sequencing to accurately sequencesegmental duplications.[40]
The T2T-CHM13 is sequenced from CHM13hTERT, a cell line from an essentially haploidhydatidiform mole. "CHM" stands for "Complete Hydatidiform Mole," and "13" is its line number. "hTERT" stands for "humanTelomerase Reverse Transcriptase". The cell line has been transfected with the TERT gene, which is responsible for maintaining telomere length and thus contributes to thecell line's immortality.[41] A hydatidiform mole contains two copies of the same parental genome, and thus is essentially haploid. This eliminates allelic variation and allows better sequencing accuracy.[40]
For much of a genome, the reference provides a good approximation of the DNA of any single individual. But in regions with highallelic diversity, such as themajor histocompatibility complex in humans and themajor urinary proteins of mice, the reference genome may differ significantly from other individuals.[43][44][45] Due to the fact that the reference genome is a "single" distinct sequence, which gives its utility as an index or locator of genomic features, there are limitations in terms of how faithfully it represents the human genome and itsvariability. Most of the initial samples used for reference genome sequencing came from people of European ancestry. In 2010, it was found that, byde novo assembling genomes from African and Asian populations with the NCBI reference genome (version NCBI36), these genomes had ~5Mb sequences that did not align against any region of the reference genome.[46]
Following projects to the Human Genome Project seek to address a deeper and more diverse characerization of the human genetic variability, which the reference genome is not able to represent. TheHapMap Project, active during the period 2002 -2010, with the purpose of creating ahaplotypes map and their most common variations among different human populations. Up to 11 populations of different ancestry were studied, such as individuals of theHan ethnic group from China,Gujaratis from India, theYoruba people from Nigeria orJapanese people, among others.[47][48][49][50] The1000 Genomes Project, carried out between 2008 and 2015, with the aim of creating a database that includes more than 95% of the variations present in the human genome and whose results can be used in studies of association with diseases (GWAS) such as diabetes, cardiovascular or autoimmune diseases. A total of 26 ethnic groups were studied in this project, expanding the scope of the HapMap project to new ethnic groups such as theMende people of Sierra Leone, theVietnamese people or theBengali people.[51][52][53][54] TheHuman Pangenome Project, which started its initial phase in 2019 with the creation of the Human Pangenome Reference Consortium, seeks to create the largest map of human genetic variability taking the results of previous studies as a starting point.[55][56]
Since the Human Genome Project was finished, multiple international projects have started, focused on assembling reference genomes for many organisms. Model organisms (e.g., zebrafish (Danio rerio), chicken (Gallus gallus),Escherichia coli etc.) are of special interest to the scientific community, as well as, for example, endangered species (e.g., Asian arowana (Scleropages formosus) or the American bison (Bison bison)). As of August 2022, the NCBI database supports 71 886 partially or completely sequenced and assembled genomes from different species, such as 676mammals, 590birds and 865fishes. Also noteworthy are the numbers of 1796insects genomes, 3747fungi, 1025plants, 33 724bacteria, 26 004virus and 2040archaea.[57] A lot of these species have annotation data associated with their reference genomes that can be publicly accessed andvisualized in genome browsers such asEnsembl andUCSC Genome Browser.[58][59]
^Hurst J, Beynon RJ, Roberts SC, Wyatt TD (October 2007).Urinary Lipocalins in Rodenta:is there a Generic Model?. Chemical Signals in Vertebrates 11. Springer New York.ISBN978-0-387-73944-1.
^Li R, Li Y, Zheng H, Luo R, Zhu H, Li Q, et al. (January 2010). "Building the sequence map of the human pan-genome".Nature Biotechnology.28 (1):57–63.doi:10.1038/nbt.1596.PMID19997067.S2CID205274447.
