How viruses work

A virus is tiny. Far smaller than one of your cells, so small you would need an electron microscope to see it. Strip it down and it is basically a bit of genetic code, its instructions, wrapped in a coat of protein.

On its own it can do nothing at all. It cannot grow. It cannot eat. It cannot even make a copy of itself. A virus just drifts until it bumps into the right kind of living cell.

Then it goes to work. It latches onto the cell, breaks in, and hands over its instructions. Those instructions trick the cell into dropping its own business and building new virus parts instead, thousands of them. The cell becomes a factory working for the invader. The new copies pour out, often bursting the cell open, and each one drifts off to find another cell to hijack.

That is why viruses make you ill, and why they spread from person to person: one cell can turn into thousands of viruses, and those go on to infect more cells. It also explains two things people mix up. Antibiotics do nothing to a virus, because those drugs are built to kill bacteria, which are living cells with their own machinery to attack. A virus has none of that. Vaccines take a different route: they show your body a harmless preview of the virus so your defences already recognise it and can shut it down fast.

Take a virus apart and there is not much to it. At the core sits the genome, the genetic instructions, written in either DNA or RNA. Around that is a protein shell called the capsid, built from many copies of a few protein shapes locked together, which is why so many viruses look like neat geometric solids. Some viruses add an outer coat of fatty membrane, a lipid envelope, stolen from the last cell they left. Studded through that envelope are proteins that stick out like spikes. The spikes on a coronavirus are the famous example, and they are the part that grabs onto the cell.

Infection runs through a set cycle. First, attachment: a viral protein has to fit a specific receptor on the cell surface, like a key meeting its lock. Then entry, where the virus or its genome gets inside, followed by uncoating, when the capsid comes apart and frees the genome. Now the hijack proper. The virus has no factory of its own, so it borrows the cell's: it feeds its genome into the host's machinery, which copies the genome and reads it to churn out viral proteins. Fresh genomes and fresh proteins snap together in assembly, and finally release. Some viruses burst the cell outright, killing it, which is the lytic route. Others bud out slowly through the membrane, wrapping themselves in it as they go.

Because attachment depends on that key-and-lock fit, a virus can only infect cells that carry the matching receptor. This is why viruses are so picky. A cold virus targets your airway lining and ignores everything else; a virus of one species often cannot touch another. Host range and tissue specificity both come down to which doors a virus's proteins can open.

Copying is not perfect, and that matters. When a cell copies a viral genome it makes mistakes, and the enzymes that copy RNA are especially sloppy, with no way to check their work. So RNA viruses like influenza and SARS-CoV-2 mutate quickly. Most mutations do nothing or break the virus, but a few change the spikes enough that immune systems no longer recognise them. That constant drift is why new variants keep appearing and why the flu shot is reformulated every year to chase the strains in circulation.

All of which raises an old question. A virus carries genes and evolves, which feels alive. But it has no cells, no metabolism, and cannot reproduce without hijacking someone else, which feels like a chemical, not a creature. Biologists still argue about it, and most settle on calling viruses something in between, active only once they are inside a living host.

Classifying by genome and strategy

The Baltimore system sorts viruses not by shape or host but by the form of their genome and the route it takes to messenger RNA, since making mRNA is the one thing every virus must do to be read by the host's ribosomes. The seven classes span double-stranded DNA viruses (herpesviruses, poxviruses), single-stranded DNA, double-stranded RNA, positive-sense single-stranded RNA whose genome acts directly as mRNA (coronaviruses, picornaviruses), negative-sense single-stranded RNA that must first be transcribed by a packaged polymerase (influenza, rabies), and the retroviruses, whose \(+\)ssRNA is reverse-transcribed into DNA by reverse transcriptase and then integrated into the host chromosome. The replication strategy, not the anatomy, defines the group.

Receptor binding and specificity

Tropism is set at the first step. A receptor-binding domain on a viral surface protein must match a host molecule with enough affinity to trigger entry. SARS-CoV-2 spike binds the ACE2 receptor; HIV gp120 engages CD4 together with a chemokine co-receptor such as CCR5. Single amino-acid changes in these domains can widen or narrow host range, and they are exactly the changes that drive spillover from animals to humans.

Mutation, quasispecies and antigenic change

RNA-dependent RNA polymerases lack proofreading, giving error rates near \(10^{-4}\) per base, so an RNA virus exists not as one sequence but as a swarm of related genomes, a quasispecies, around a consensus. That standing diversity is raw material for immune escape. Antigenic drift is the slow accumulation of point mutations in surface proteins; antigenic shift is the abrupt reassortment of whole genome segments when two influenza strains co-infect a cell, and it is the shift events that seed pandemics.

Latency and integration

Not every infection is a fast burn. Herpesviruses persist as quiet episomes in neurons and reactivate under stress. Retroviruses like HIV integrate a DNA provirus into the host genome, where it can sit transcriptionally silent, invisible to both the immune system and to drugs that only hit replicating virus. This latent reservoir is the central obstacle to curing HIV rather than merely suppressing it.

Why antivirals are hard

A virus runs most of its life on borrowed host machinery, so there are few targets that are viral and not also human. Effective antivirals hit the steps that are genuinely the virus's own: entry inhibitors block receptor engagement or fusion, nucleoside analogues and polymerase inhibitors jam genome copying, protease inhibitors stop the cleavage of viral polyproteins, and integrase inhibitors block proviral insertion. Because each drug is tuned to one virus's enzymes, and because high mutation rates breed resistance, broad-spectrum antivirals as reliable as broad-spectrum antibiotics remain rare.

Bacteriophages

The most abundant viruses on Earth do not infect us at all. Bacteriophages prey on bacteria, shaping microbial populations across the oceans and the gut, and their study handed molecular biology many of its founding tools. They are now being revisited as phage therapy against antibiotic-resistant infections, and their defence systems gave us CRISPR.