How to Read a Reference Genome
Why This Concept Matters
Genome-analysis results often contain positions such as chr7:.... A chromosome name and a number are not a complete address, however. You must know which genome assembly defines the coordinate before you can locate the same sequence region or variant.
A reference genome is more than a โrepresentative human DNA sequence.โ It is a sequence-based coordinate system used to align sequencing reads, record variants and gene features, and exchange results. If the reference is not fixed at the beginning, coordinates, annotations, and analysis resources can become incompatible.
A Reference Is Not Just One FASTA File
| Layer | What to record | What can change |
|---|---|---|
| assembly | GRCh37, GRCh38, T2T-CHM13, and so on | the same coordinate number may identify different sequence |
| assembly release | major/patch and accession/version | the exact included sequences become ambiguous |
| reference bundle | primary, alternate, decoy, and other FASTA components | read mapping and variant calls can differ |
| annotation | provider, release, and GTF/GFF | gene and transcript boundaries can differ |
| analysis resources | known sites, intervals, and indexes | contig names or coordinates may no longer match |
โWe used GRCh38โ is necessary information, but it is not sufficient. Record the assembly accession, the versions of the FASTA and supporting files, and the annotation release.
GRCh37, hg19, GRCh38, and hg38
GRCh37 and GRCh38 are Genome Reference Consortium assembly names. hg19 and hg38 are the widely used corresponding UCSC names. GRCh19 is not an official assembly name; it usually results from mixing GRCh37 with hg19.
The current GRC human major assembly is GRCh38, with GRCh38.p14 as its latest patch release. A patch release adds patch sequences without changing the chromosome coordinates of the major assembly. A major change from GRCh37 to GRCh38, by contrast, requires coordinates and sequence context to be checked again.
Can We Change Assemblies by Replacing the Address?
Not always. A conserved region with one unique correspondence may convert cleanly. Repeats, segmental duplications, structural differences, and newly added regions can instead produce unmapped or multi-mapped features, changed intervals or strands, and reference-allele mismatches.
Tools such as UCSC LiftOver and Ensembl Assembly Converter project coordinates using assembly alignments. A successful conversion does not guarantee biological identity. Validate the mapping, interval, strand, allele, and target annotation. For complex regions, realigning the original reads to the target reference may be more defensible than projecting coordinates.
confirm source assembly and reference allele
โ
fix target assembly and conversion method
โ
review unmapped / multi-mapped / changed intervals
โ
recheck target allele and annotation releaseWhat Does KOREF Add?
The 2016 KOREF study presented KOREF_S, based on one donor, and KOREF_C, which incorporated common variants from 40 high-depth Korean genomes. Its purpose was not to replace GRCh38 with a single Korean standard, but to assess whether population-relevant sequence and variant representation could improve Korean and East Asian genome analysis.
The later KOREF_S1 is a chromosome-level assembly built with long reads, Hi-C, and parental information. โKOREFโ must therefore be accompanied by an exact release. It is neither a single normal genome for all Koreans nor a clinical standard.
Why Is GRCh38 Still Used After T2T and Pangenomes?
T2T-CHM13 substantially extends gaps and complex regions missing from earlier references. The Human Pangenome Reference represents phased assemblies from multiple people and their relationships, reducing variation missed by a single linear reference. A T2T single-assembly reference and a population-diversity pangenome are not the same concept.
This does not automatically invalidate GRCh38-based results. Annotation, population databases, assay designs, clinical knowledge, and software ecosystems remain tied to specific references. The key is not to memorize one โnewest reference,โ but to state the scientific question, exact release, reference bundle, annotation, and conversion validation.
Common Misunderstandings
- A reference genome is not one person's perfect or universally normal genome.
- GRC patches preserve major chromosome coordinates, but the bundle still needs to be recorded.
- Successful coordinate conversion does not establish complete variant identity.
- A pangenome is not simply the next numbered version of GRCh38.
Evidence Boundary
This Concept explains reference selection and reporting. It does not provide instructions for a specific assay, clinical variant reinterpretation, or executable conversion commands. A practical tutorial must separately lock the reference bundle, annotation release, and checksums.