IELTS Academic Reading Practice Test 376

On 26 June 2000, two men who had spent the previous two years as bitter rivals stood beside the American president to announce a shared achievement. One led an international consortium of publicly funded laboratories; the other ran a private company. Between them, they said, they had assembled a working draft of the human genome: the complete set of instructions, written in DNA, for building a human being. The British prime minister took part from London. The draft was published the following February, in rival papers in the journals Nature and Science.

The idea of reading the entire human genome had been discussed since the mid-1980s, when faster sequencing made such a project conceivable. In 1988 the United States created an office within the National Institutes of Health to coordinate the work, with James Watson, co-discoverer of the structure of DNA, as its first director. The Human Genome Project itself began in 1990, with a plan to run for fifteen years, and its budget was usually quoted as three billion dollars. Laboratories in Britain, France, Germany and Japan joined the American effort, and China became a partner in 1999.

Reading DNA in the 1990s was slow work. The standard technique, devised by Frederick Sanger in the 1970s, copies a strand of DNA and then interrupts the copying at random points, producing fragments of every possible length, each ending in a labelled base. Sorted by size, the fragments can then be read off in order. The method is accurate, but it yields only a few hundred letters at a time, so the genome had to be broken into large overlapping pieces, each carried inside a bacterium, before sequencing could begin. By 1998, eight years into the programme, only a small fraction of the sequence had been completed.

Then came a shock. In 1998 a biologist named Craig Venter founded a company, Celera Genomics, and announced that he would finish the genome in three years. Rather than mapping large pieces first, Celera proposed to shatter the whole genome into millions of small fragments, read them all at once and let computers work out how they overlapped — an approach known as whole-genome shotgun sequencing. Most academic geneticists dismissed the idea as unworkable for a sequence so full of repeated DNA, but Venter's machines delivered data at an extraordinary rate, and the public consortium was forced to speed up its programme.

The rivalry raised another question: who would own the sequence? In 1996, at a meeting in Bermuda, the publicly funded laboratories had agreed that every fragment of sequence should be placed in a free public database within twenty-four hours of being read. The rule was radical, and it shaped the project's reputation, because researchers anywhere could use the data immediately. Celera's sequence, by contrast, came with conditions attached for anyone who wanted it for commercial purposes, though it did deposit a version in a public archive. The dispute was overtaken by events: the consortium's policy established open release as the norm for large biology projects.

The results confounded expectations. The human genome turned out to contain about three billion base pairs, the chemical letters of DNA, arranged along twenty-three pairs of chromosomes. But the number of protein-coding genes was startlingly small. Before the project, biologists had commonly estimated that a human being would need around 100,000 genes to account for the complexity of the body; the final count was closer to 20,000, not much more than a nematode worm's. Only about 1.5 per cent of the sequence consists of instructions for making proteins. The rest includes vast stretches of repeated DNA and an elaborate system of switches that determines when and where genes are active.

A finished sequence was announced in April 2003, shortly after the fiftieth anniversary of Watson and Crick's description of the double helix. It covered about 99 per cent of the gene-containing portion of the genome. Even so, several hundred gaps remained, mostly in regions where DNA is highly repetitive or where it lies tightly coiled at the ends and centres of chromosomes. These regions resisted every technique available at the time; they were not closed until 2022, when a consortium using long-read methods published a truly complete sequence.

From the beginning, the project's advocates had promised medical benefits, and genes responsible for single-gene disorders were identified during these years: the gene for cystic fibrosis in 1989, the gene for Huntington's disease in 1993, and the two principal genes linked to inherited breast cancer, BRCA1 and BRCA2, in 1994 and 1995. Yet the common diseases that fill hospital wards turned out to be influenced by hundreds or thousands of variants, each with a tiny effect, interacting with environment and chance. New treatments have arrived more slowly than the publicity implied, and the clearest gain has been diagnostic rather than therapeutic.

The project's most durable legacy may therefore be infrastructure rather than medicine. It produced a reference sequence against which every later study is measured, freely available maps and tools, and a generation of researchers accustomed to sharing data. It also broke new ground by funding work on the ethical, legal and social implications of genetics, a decision that made questions of insurance, privacy and consent part of the science rather than an afterthought. And it made sequencing cheap: a task that cost three billion dollars in 2003 can now be done for a few hundred.

A genetic test is an exercise in comparison. The laboratory reads the letters of a person's DNA and sets them against a reference sequence, reporting every point at which they differ. Human beings are almost identical in this respect: across three billion base pairs, any two people vary at roughly one position in a thousand. Most of those differences have no effect at all. Some alter the behaviour of a gene, and a small minority either cause a disorder outright or make it more likely. A test may examine a single letter, a single gene, the protein-coding portion of the genome — the exome, about one per cent of the whole — or the entire sequence.

Every test begins with a sample, generally blood or saliva, from which DNA is extracted and purified. The difficulty is quantity. A cell contains only one copy of the genome, and most reading methods require many thousands of copies of the region of interest. The solution is the polymerase chain reaction, devised by Kary Mullis in 1983. The sample is heated so that the two strands of the double helix separate, short synthetic primers bracket the target, and an enzyme copies the DNA between them. Each cycle doubles the amount, so that thirty cycles turn a single copy into more than a billion.

For a single gene, many laboratories still rely on the method Frederick Sanger developed in the 1970s. It copies the DNA but interrupts the copying at random points by inserting a modified base that carries a fluorescent dye, so that fragments of every length are produced, each ending in a labelled letter. Electrophoresis sorts the fragments by size, and a laser reads the dye on each as it passes. The technique is accurate and reads stretches of 700 to 900 letters, but it handles one fragment at a time and is far too slow and expensive for a whole genome.

That changed from about 2005, when machines appeared that could read millions of fragments simultaneously. DNA is broken into short pieces, and adapters are attached to their ends so that they bind to a surface. Each piece is copied in place, generating a cluster of identical molecules whose signal can be detected, and a camera records the colour of the base added at each step. The output is enormous — hundreds of billions of letters per run — but the reads are short, typically 100 to 300 letters. Reading an entire human genome, once a multi-billion-dollar undertaking, now costs a few hundred dollars, a decline steeper than that of computer chips.

The short reads must then be put back together. Software aligns each one to the reference genome, using overlaps and quality scores, and flags the positions where a sample departs from the standard. Repetitive stretches remain difficult, because a read that matches the reference in several places can be assigned to the wrong one. Departures take several forms: a single letter exchanged for another; a small insertion or deletion; a gene present in extra copies or missing entirely; or a short sequence repeated too many times, as in the gene implicated in Huntington's disease. The result is a list of variants, sometimes running to millions of entries in a whole-genome test.

A list is not a diagnosis. Most variants are harmless, and a few are well documented causes of disease, but a large and awkward middle category remains: variants of uncertain significance, whose effect is unknown. Laboratories grade each finding against published guidelines and consult databases in which other laboratories have recorded what they found and what happened to the patient. Interpretation is complicated by the reference itself. Much of the standard sequence comes from a single donor, and the databases hold far more information about people of European ancestry than about others, so a patient from a less-studied population is more likely to receive an uncertain result.

Testing is now routine in several parts of medicine. It confirms rare inherited disorders and identifies the exact change responsible. In cancer care it can show which drug is likely to work against a particular tumour. It guides prescribing: the HIV drug abacavir can cause a severe reaction in people who carry a particular immune-system variant, and testing before treatment is now recommended. Screening of newborn babies for a small number of treatable conditions has been carried out for decades, and a newer test can detect chromosomal abnormalities in a fetus from fetal DNA circulating in the mother's blood, without any risk to the pregnancy.

What testing cannot yet do is predict common disease with any confidence. For most such conditions, hundreds or thousands of variants each contribute a very small increase in risk, and a polygenic score that adds them up gives a probability, not a prognosis. Such scores often perform less well in populations other than the one in which they were derived. A result may also reveal information the patient did not seek, such as misattributed paternity or a risk of a condition for which nothing can be done. For these reasons, professional bodies recommend that significant findings be discussed with a genetic counsellor, whose task is to explain what is and is not known rather than to make the decision.

Imagine a woman in her thirties whose mother and aunt both developed breast cancer at an early age. A test shows that she carries a variant in a gene called BRCA1 that raises her own risk considerably. Should she have to tell an insurer before taking out a life policy, and may the insurer charge her more, or refuse her altogether, because of what the test found? Every country with a private insurance market has had to answer some version of this question, and the answers differ sharply. The argument is not really about genes. It is about what insurance is for, and about how much inequality of information a market can bear.

The industry's starting point is that insurance is a mechanism for sharing risk, not a promise that everyone will pay the same price. Companies sort applicants by age, smoking, weight, occupation and medical history, and charge more to those who are more likely to claim. If that seems hard, the argument runs, it is no harder than the lottery of illness itself, and any classification that reflects real differences in risk keeps premiums lower for the majority. Insurers also point out that family history is already on the application form: a woman with two affected relatives will be treated as a higher risk whether or not she has been tested. On this view, genetic information is not a special category at all, and the demand that it be treated differently is a form of superstition rather than principle.

The industry's strongest argument is technical. Insurance works only if the applicant and the insurer know roughly the same about the applicant's health. If customers can learn privately that they are likely to fall ill and the insurer cannot, those at high risk will buy larger policies and those at low risk will buy none. Premiums must then rise, driving out more of the healthy, and the spiral can end in a market that offers cover only to those who do not need it. Economists call this adverse selection, and it is the reason why insurance markets are regulated at all.

Yet the evidence does not suggest that genetic tests would destabilise insurance. For a few rare single-gene disorders, such as Huntington's disease, a predictive test is close to a diagnosis, but such conditions are, fortunately, too uncommon to matter to an insurer's accounts. For the diseases that account for most claims — cancer, heart disease, diabetes — common genetic variants add very little to what an insurer already knows from age and family history. A polygenic score that moves a person's risk from average to slightly above average scarcely justifies a large increase in premium. Genes are not destiny either: identical twins do not necessarily develop the same diseases, and much of the variation in common illness remains unexplained by any test on the market.

Against any gain in actuarial accuracy must be set the consequences of disclosure. A test result is unlike a blood pressure reading: it does not change when you lose weight or stop smoking, it is shared with relatives who never chose to be tested, and it may predict events decades ahead. Once it is in an insurance file it cannot be withdrawn. Studies of people at risk of inherited conditions have reported that some decline testing altogether, or decline to take part in research, rather than create a record they fear. That is a poor outcome for the individual, and a poor one for medicine.

Regulation has followed national instincts. In the United States, the Genetic Information Nondiscrimination Act of 2008 bars health insurers and employers from using genetic information, but it says nothing about life insurance, disability insurance or long-term care, the products where the sums at stake are largest. In the United Kingdom, insurers have agreed with the government not to ask applicants for predictive test results, an arrangement first made in 2001 and renewed several times since, with a single narrow exception for Huntington's disease and only for very large policies. Several European countries go further and forbid insurers from using genetic information at all, while others have no specific rule.

My own view is that the balance of argument favours a statutory ban on the use of predictive genetic results by insurers, not merely a voluntary agreement. Voluntary arrangements can be revised when insurers press for change, and they give an applicant no means of challenging a decision. A law should cover life and disability insurance as well as health cover, since a refused policy can mean a refused mortgage, and it should protect the results of research as well as of clinical tests, because the two are increasingly hard to separate. None of this requires insurers to abandon underwriting. Within a legal ban on predictive results, they would remain free to use age, family history, and the results of tests already reflected in a diagnosis or a patient's medical records.

This is not to deny that the industry's concern is real, or that a ban costs something. If applicants who know themselves to be at high risk buy cover that the insurer cannot price, the cost is ultimately carried by other policyholders, many of whom are less well off than the applicant. But the sums involved appear small, while the harm to a family denied cover at the moment it is most needed is severe and lasting, and it falls on people who have done nothing except seek information about their own health. Where the potential loss is modest and widely spread but the individual loss is concentrated, the argument for regulation is strong. Insurance exists to spread misfortune, not to sort the fortunate from the rest.