Reference Genomes and Coordinate Systems

  • Reference genomes and coordinate systems are fundamental concepts in genomics and bioinformatics. A reference genome provides a standard DNA sequence against which sequencing reads, genetic variants, genes, regulatory elements, and other genomic features can be identified and compared. A coordinate system provides a consistent way to describe the positions and boundaries of these features within the reference sequence. Together, they allow researchers to locate biological information precisely, integrate data from different experiments and databases, and reproduce genomic analyses.
  • A reference genome is a representative assembled DNA sequence for an organism or a particular individual. It is commonly organized into chromosomes, unplaced scaffolds, contigs, and other sequence components, depending on the quality and completeness of the assembly. Reference genomes are developed using sequencing technologies, genome assembly methods, and additional evidence such as genetic maps, physical maps, and chromosome-conformation data. They may represent a haploid sequence, a selected haplotype, a consensus sequence, or a more complex representation of genomic diversity. A reference genome should therefore not be regarded as a perfect or universal representation of every individual belonging to a species.
  • Reference genomes are essential for many bioinformatics applications, including read mapping, variant calling, gene annotation, comparative genomics, genome-wide association studies, transcriptomics, and evolutionary analysis. When sequencing reads are aligned against a reference genome, their positions can be expressed using genomic coordinates. These coordinates allow researchers to identify the genomic regions from which reads originate and connect sequencing data with annotated genes, transcripts, regulatory elements, and known genetic variants.
  • Several organizations maintain reference genome resources. The Genome Reference Consortium develops and maintains important human reference assemblies. NCBI provides genome assemblies through its Assembly and RefSeq resources, while the European Nucleotide Archive and other International Nucleotide Sequence Database Collaboration partners archive submitted sequence data. Ensembl provides genome assemblies and annotations for numerous organisms. Organism-specific databases also maintain reference genomes for agricultural species, plants, model organisms, and microorganisms. Examples include resources for mouse, zebrafish, Arabidopsis, cattle, chicken, and many other species.
  • A genome assembly is generally identified by an assembly name, accession, and version. In NCBI Assembly, accessions commonly use prefixes such as GCA_ for GenBank assemblies and GCF_ for RefSeq assemblies, followed by a numeric identifier and version. The version indicates a particular release of the assembly record. A new assembly version may reflect changes to the sequence or its organization, whereas annotation updates may occur independently of the underlying assembly sequence. Researchers should therefore distinguish between the assembly version and the annotation release when reporting genomic coordinates.
  • Reference assemblies may be classified according to their completeness and chromosome organization. A complete or near-complete chromosome-level assembly contains sequences organized into chromosomes, although some gaps or unresolved regions may remain. Scaffold-level assemblies contain groups of ordered and oriented contigs, while contig-level assemblies consist of continuous sequence fragments that may not yet be connected into larger structures. Telomere-to-telomere assemblies aim to provide continuous chromosome sequences from one telomere to the other, but such completeness is not available for every organism or every chromosome. Assembly quality and completeness can influence the reliability of genomic coordinates and downstream analyses.
  • A reference genome is often distributed in FASTA format, with each chromosome or sequence component represented by a separate record. The FASTA header contains an identifier and may include descriptive information. The sequence itself is represented by nucleotide characters, including ambiguity codes where appropriate. For example:

>chr1 Example reference chromosome 1

ATGCGTACGTTAGCTAGCTAGCTAGCTAGC

>chr2 Example reference chromosome 2

CGTAGCTAGGCTAACGTTAGCTAGCTAGCTA

  • This is a simplified illustrative FASTA file rather than a complete biological reference genome. The sequence identifiers in the headers are important because they are used by alignment files, annotation files, and variant records to identify genomic sequences. A reference sequence may also include mitochondrial DNA, unplaced scaffolds, alternate loci, decoy sequences, and other components. These additional sequences can affect mapping and coordinate interpretation.
  • A genomic coordinate identifies a position or interval within a particular reference sequence. It normally requires at least the reference assembly, chromosome or sequence identifier, start position, and, where relevant, end position. For example, a genomic interval may be described as chromosome 1, positions 100,000 to 100,500. This interval is meaningful only when the reference assembly and coordinate convention are known. The same numerical coordinates may refer to different DNA sequences or biological features in different genome assemblies.
  • One important distinction in genomic coordinate systems is the difference between zero-based and one-based indexing. In a one-based coordinate system, the first nucleotide of a sequence is assigned position 1. In a zero-based coordinate system, the first nucleotide is assigned position 0. These conventions can produce different numerical representations for the same genomic interval. For example, the first 10 bases of a chromosome are represented as positions 1–10 in a one-based, closed interval system, but as positions 0–10 in a zero-based, half-open interval system.
  • A one-based, closed interval includes both the start and end positions. Its length is calculated as:

Length = End – Start + 1

  • A zero-based, half-open interval includes the start position but excludes the end position. Its length is calculated as:

Length = End – Start

  • For example, the genomic interval covering the first 10 bases can be represented in two ways:

One-based, closed:       1-10

Zero-based, half-open:   0-10

  • Both representations describe the same 10 nucleotide positions. Confusing these systems can cause off-by-one errors, which may lead to incorrect extraction of DNA sequences, incorrect overlap calculations, and inaccurate annotation of genomic features.
  • Different bioinformatics file formats use different coordinate conventions. GFF3 and GTF generally use one-based, closed coordinates. VCF uses one-based positions for variant records. BED uses zero-based, half-open coordinates. SAM alignment positions are one-based, while BAM stores the corresponding alignment position internally in a zero-based representation. Pileup and other output formats may have their own conventions. Researchers must check the specification of each format rather than assuming that all genomic files use the same indexing system.
  • The BED format is widely used for representing genomic intervals. A basic BED record contains a chromosome identifier, a zero-based start coordinate, and an exclusive end coordinate. For example:

chr1    0    10

chr1    100  200

chr2    500  750

  • The first record describes the first 10 bases of chromosome 1, while the second describes a region of 100 bases. BED files can include additional columns for feature names, scores, strand information, and graphical display properties. They are commonly used for gene regions, transcription factor binding sites, sequencing peaks, conserved elements, and other genomic intervals.
  • GFF3 and GTF are annotation formats used to describe genomic features such as genes, transcripts, exons, coding sequences, and regulatory regions. Their coordinates are generally one-based and inclusive. A simplified GFF3 record looks like this:

##gff-version 3

chr1    example    gene    101    500    .    +    .    ID=gene001

chr1    example    exon    101    200    .    +    .    ID=exon001;Parent=gene001

chr1    example    exon    301    500    .    +    .    ID=exon002;Parent=gene001

  • In this example, the gene extends from position 101 to position 500 on the positive strand of chromosome 1. Its two exons occupy positions 101–200 and 301–500. The coordinates are illustrative and do not correspond to a specific organism’s annotated genome. GFF3 and GTF files also contain attributes that describe relationships between genes, transcripts, and their component features.
  • The VCF format is used to represent genetic variants, including single-nucleotide variants, insertions, deletions, and other forms of genomic variation. Its POS field uses one-based coordinates to identify the reference position associated with each variant record. For example:

##fileformat=VCFv4.3

#CHROM  POS  ID  REF  ALT  QUAL  FILTER  INFO

chr1    101  .   A    G    60    PASS    .

chr1    205  .   AT   A    50    PASS    .

  • The first record describes a substitution at position 101, while the second illustrates a deletion represented using a reference allele of AT and an alternate allele of A. Indels in VCF are represented using an anchoring base according to the VCF specification. Therefore, the POS field should not be interpreted simply as the first deleted or inserted nucleotide without considering the complete REF and ALT representation.
  • The SAM and BAM formats also contain genomic coordinates. In SAM, the alignment position is stored in the POS field using one-based coordinates. The alignment location is identified by the reference sequence name and position, while the CIGAR string describes how the read aligns to the reference. Insertions, deletions, clipping, and skipped reference regions can affect the relationship between read positions and reference coordinates. BAM uses an internal zero-based coordinate representation, although software generally presents positions according to the relevant file-format conventions when converting or displaying records.
  • A major challenge in genomic analysis is maintaining consistent reference sequence identifiers across files. One resource may use chr1, while another uses 1. Similarly, mitochondrial sequence identifiers may appear as chrM, MT, or another accession-based name. These differences can prevent software from matching alignments, annotations, and variants even when the underlying sequences are identical. Researchers should use compatible reference files or carefully convert identifiers while preserving the identity of each sequence.
  • Reference genome assemblies may contain alternate loci, unplaced scaffolds, unlocalized scaffolds, decoy sequences, and haplotype-specific sequences. Alternate loci represent alternative sequence representations of particular genomic regions, while unplaced scaffolds have not been assigned to a specific chromosome. Unlocalized scaffolds are assigned to a chromosome but have an uncertain position or orientation. Decoy sequences may be included to improve the handling of reads that originate from sequences absent or difficult to represent in the primary assembly. The exact composition of a reference package can therefore affect alignment results and coordinate interpretation.
  • The primary assembly is often used as the main coordinate framework, but some genomic regions contain substantial variation between individuals. Alternative haplotypes and alternate loci can represent sequence differences that are not adequately captured by a single linear reference. In such cases, reads may align to multiple related regions, and variant representation may depend on the reference sequence selected. Researchers should document whether they used a primary assembly, an alternate-aware reference, or a specialized reference package containing additional sequences.
  • Reference genome versions are particularly important when comparing data from different studies. An annotation created for one assembly may not match coordinates from another assembly because insertions, deletions, rearrangements, corrections, or changes in chromosome organization can shift genomic positions. For example, a gene located at one coordinate in an older assembly may be located at a different coordinate in a newer assembly. This does not necessarily mean that the gene has moved biologically; it may reflect changes in the reference sequence or assembly structure.
  • Genome coordinate conversion, commonly called liftover, is used to translate genomic positions between reference assemblies. It relies on alignment or mapping relationships between the old and new assemblies. Tools such as UCSC liftOver and CrossMap can convert genomic intervals between supported assemblies. NCBI also provides resources for assembly conversion and sequence correspondence. However, not every interval can be converted successfully. Regions may have been deleted, rearranged, duplicated, or assembled differently, and some intervals may map to multiple locations. A converted coordinate should therefore be checked for mapping quality, uniqueness, and biological consistency.
  • An illustrative liftover command using the UCSC liftOver tool is:

# Convert BED intervals from one assembly to another

liftOver input.bed oldToNew.over.chain output.bed unmapped.bed

  • The chain file describes the correspondence between the source and target assemblies. It must match the correct direction of conversion and the intended assemblies. The output file contains successfully converted intervals, while the unmapped file contains intervals that could not be converted. Liftover is not a substitute for reanalysing raw sequencing data against the appropriate reference when accurate alignment or variant calling is required.
  • Genomic coordinate systems also include strand information. DNA sequences have two complementary strands, commonly represented as the positive (+) and negative (−) strands. A gene can be located on either strand of a chromosome. For genes on the positive strand, transcription generally proceeds in the direction of increasing genomic coordinates. For genes on the negative strand, transcription proceeds in the direction of decreasing genomic coordinates. Consequently, the order of exons in transcript orientation may differ from their numerical order in genomic coordinates.
  • Strand information is especially important when extracting sequences, interpreting gene structures, identifying splice junctions, and analysing regulatory elements. A sequence extracted from the negative strand may need to be reverse-complemented to obtain the sequence in the gene’s transcriptional orientation. Bioinformatics tools generally provide options to handle strand information, but researchers must ensure that the correct strand convention is used.
  • Transcript coordinates and genomic coordinates are related but are not interchangeable. A gene may contain multiple exons separated by introns, while a mature transcript contains the joined exon sequences. Therefore, a position within a transcript does not necessarily correspond to the same numerical position in the genome. For example, transcript position 100 may lie within an exon located much farther along a chromosome. Alternative splicing can also produce multiple transcripts from the same gene, each with a different exon composition and transcript-coordinate system.
  • Coding-sequence coordinates introduce another layer of complexity. The coding sequence, or CDS, represents the part of a transcript translated into protein, generally beginning at a start codon and ending at a stop codon. The CDS may consist of several separate genomic intervals because introns are removed during RNA processing. In GFF3 and GTF annotations, the CDS is often represented by multiple records associated with a transcript. The phase or frame information indicates how the coding sequence continues across exon boundaries. Correct interpretation is important for translating sequences and identifying coding variants.
  • Coordinate systems are also used in comparative genomics and population studies. Researchers may compare genes, variants, conserved regions, and structural features across different individuals or species. However, genomic coordinates are specific to a reference sequence and cannot automatically be compared across species or assemblies. Orthology relationships, sequence alignments, synteny, and assembly-to-assembly mappings are needed to establish correspondence between genomic regions. A gene on chromosome 1 in one species may have homologous sequences on different chromosomes or in rearranged regions in another species.
  • Several tools support genomic coordinate manipulation and interval analysis. BEDTools is widely used for comparing, intersecting, merging, sorting, and extracting genomic intervals. Its functions include identifying overlapping features, calculating coverage, and finding distances between genomic regions. BEDOPS provides efficient operations for large genomic interval datasets. UCSC utilities support coordinate conversion and genome-browser workflows, while CrossMap supports conversion of genomic coordinates and related annotations between reference assemblies. SAMtools and bcftools are commonly used to manipulate alignment and variant files, while tabix supports indexed access to genomic interval data.
  • For example, BEDTools can identify overlapping intervals between two BED files:

# Find overlaps between two BED files

bedtools intersect -a genes.bed -b variants.bed

# Report the number of overlapping bases

bedtools intersect -a genes.bed -b variants.bed -wo

# Merge overlapping or nearby intervals

bedtools merge -i regions.sorted.bed

  • The input files should be sorted and use compatible reference sequence identifiers and coordinate conventions. BEDTools expects BED-style coordinates for BED input, so converting annotation or variant data into BED requires careful coordinate handling.
  • Python libraries can also be used to work with genomic coordinates. Biopython provides sequence and annotation-related functionality, while libraries such as PyRanges and pyranges support efficient genomic interval operations. The following simplified Python example illustrates the conversion of a one-based, closed interval into a zero-based, half-open interval:

def one_based_to_bed(start, end):

    “””

    Convert a one-based, closed interval

    to a zero-based, half-open interval.

    “””

    if start < 1 or end < start:

        raise ValueError(“Invalid one-based interval”)

    return start – 1, end

start, end = 101, 200

bed_start, bed_end = one_based_to_bed(start, end)

print(bed_start, bed_end)

The output is:

100 200

  • The original interval covers 100 bases, and the converted BED interval also covers 100 bases. Such simple conversions are useful, but real-world workflows must also handle chromosome identifiers, strand information, multiple intervals, and special cases such as insertions and structural variants.
  • Genome browsers provide visual interfaces for exploring reference genomes and their annotations. The UCSC Genome Browser, Ensembl Genome Browser, and NCBI Genome Data Viewer allow researchers to navigate chromosomes, inspect genes, examine sequence variants, and compare genomic features. They also provide links to external databases and analysis resources. Genome browsers are particularly useful for checking whether genomic coordinates correspond to the expected gene, transcript, regulatory region, or variant. However, the selected genome assembly must always be checked because browser coordinates may differ between assembly versions.
  • Reference genome choice is a major consideration in studies involving genetically diverse populations, agricultural species, and non-model organisms. A reference assembled from one individual may not represent all genetic variants present in a population. This can lead to reference bias during read mapping and variant discovery. Pangenome approaches address some of these limitations by representing genomic sequences from multiple individuals or populations. Graph-based genome representations can include alternative paths corresponding to different haplotypes, insertions, deletions, and structural variants. These approaches provide a more flexible framework than a single linear reference, although their coordinate systems and analysis methods can be more complex.
  • The quality of a reference genome also affects coordinate reliability. Assembly errors, gaps, collapsed repeats, incorrect scaffold placement, and unresolved structural variation can cause coordinates to represent an inaccurate genomic structure. Annotation errors can further affect the interpretation of gene boundaries, exon positions, and regulatory regions. Researchers should therefore consider assembly quality metrics, annotation evidence, and the biological characteristics of the organism before selecting a reference genome.
  • For reproducible genomic analysis, researchers should record the organism, reference assembly accession and version, source database, annotation release, reference FASTA checksum, chromosome naming convention, and coordinate system used. They should also document any masking, sequence removal, or modification of the reference. When converting coordinates between assemblies or file formats, the conversion method and software version should be reported. These details are particularly important when genomic coordinates are shared in publications, supplementary files, databases, or collaborative research projects.
  • Reference genomes and coordinate systems provide the framework for organizing and interpreting genomic information. A reference genome supplies the sequence against which biological data are analysed, while a coordinate system defines the positions and boundaries of genomic features. Understanding assembly versions, chromosome identifiers, strand orientation, coordinate conventions, annotation formats, and genome liftover is essential for accurate genomic analysis. Consistent use of these principles helps prevent coordinate errors, supports data interoperability, and improves the reliability and reproducibility of bioinformatics research.
Author:

Leave a Reply

Your email address will not be published. Required fields are marked *