Genome, genes & chromatin

DNA: the molecule

DNA (deoxyribonucleic acid) is a polymer of four nucleotides. Each nucleotide has a deoxyribose sugar, a phosphate and one of four bases: the purines adenine (A) and guanine (G), and the pyrimidines cytosine (C) and thymine (T). Two strands wind into a right-handed double helix and pair by hydrogen bonds: A with T (two bonds) and G with C (three bonds). Because G–C pairs are stronger, the GC content of a region affects its melting temperature, how well it is captured and amplified in library preparation, and therefore sequencing coverage. GC bias is a recurring source of technical artefacts in nearly every sequencing-based omics assay.

The strands are antiparallel: one runs 5′→3′, the other 3′→5′. Polymerases only synthesise in the 5′→3′ direction, so by convention sequences are written 5′→3′. A gene can sit on either strand, which is why annotation files (GTF/GFF) carry a + or − strand field and why "strand-specific" RNA-seq libraries matter: they tell you which strand an RNA was transcribed from, so overlapping antisense genes can be separated.

How the human genome is organised

The two reference assemblies you will meet are GRCh37/hg19 (2009, still used by many clinical labs) and GRCh38/hg38 (2013, current standard), plus the complete T2T-CHM13 (2022) and the Human Pangenome Reference (a graph of many haplotypes, first release 2023). Coordinates differ between builds, so mixing them is one of the most common silent errors. Use liftOver or CrossMap to convert, and always record the build.

Anatomy of a protein-coding gene

Reading 5′→3′ along the coding strand, a typical eukaryotic gene contains:

• Promoter: the region around the transcription start site (TSS) where RNA polymerase II and general transcription factors assemble. Core promoter elements include the TATA box (~25–30 bp upstream, in a minority of genes), the initiator and downstream promoter element. About 60–70% of human promoters sit in CpG islands, regions of high CpG density that are usually unmethylated when the gene is active.
• Enhancers and silencers: distal regulatory elements, sometimes hundreds of kilobases away, bound by sequence-specific transcription factors. They contact promoters through 3D chromatin looping. Most disease-associated GWAS variants fall in enhancers rather than in coding sequence.
• 5′ untranslated region (5′UTR): transcribed but not translated; contains elements controlling ribosome recruitment, including upstream open reading frames (uORFs).
• Exons and introns: exons are retained in mature mRNA; introns are removed by splicing. Human genes average ~9–10 exons; introns are on average much longer than exons.
• Coding sequence (CDS): from the start codon (AUG) to a stop codon (UAA, UAG, UGA), read in triplets.
• 3′UTR and polyadenylation signal: the AAUAAA hexamer directs cleavage and poly(A) tail addition. 3′UTRs carry microRNA binding sites and RNA-binding protein motifs that control stability, localisation and translation. Alternative polyadenylation produces isoforms with longer or shorter 3′UTRs.

Chromatin: packaging that controls access

DNA wraps ~147 bp at a time around a nucleosome, an octamer of histones H2A, H2B, H3 and H4. Nucleosomes are separated by linker DNA (bound by H1), giving the "beads on a string" fibre with a ~200 bp periodicity. This periodicity is visible directly in ATAC-seq fragment-length distributions and in cell-free DNA fragment sizes (~167 bp peak), which is why both are used to infer nucleosome positioning.

Chromatin exists in two broad states:

At larger scales the genome folds into loops (often anchored by CTCF and cohesin), topologically associating domains (TADs) of roughly 100 kb–1 Mb that preferentially self-interact, and A/B compartments that separate active from inactive chromatin. These are measured with chromosome conformation capture (Hi-C, Micro-C) and covered in the epigenomics chapter. Disrupting a TAD boundary can let an enhancer activate the wrong gene; this mechanism explains some limb malformations and cancers.

Cells, tissues and why context matters

Almost every cell in a person carries the same genome, yet a neuron and a hepatocyte use very different sets of genes. Cell identity is established by transcription factors and maintained by the epigenome. For analysis this has three consequences:

1. Bulk measurements average over cell types. A change in a bulk tissue profile can mean that cells changed state, or only that the cell-type proportions changed. Deconvolution and single-cell methods exist to separate these.
2. Regulatory effects are context-specific. A variant may affect expression only in one cell type, developmental stage or stimulus condition, which is a central difficulty when interpreting GWAS hits.
3. Somatic variation exists. Cells acquire mutations as they divide. Cancer is the extreme case, but normal tissues are also mosaics of clones, which matters for low-frequency variant detection.