Backroom Boys

by Francis Spufford

  The scope for coaxing lies in the fact that DNA is designed to reproduce itself. Every time a cell divides, the nucleus splits into two, and the DNA inside the nucleus does the same by unzipping down the middle of the double helix. Each of the two separated strands of DNA then grows back into a complete double helix again, with the help of another molecular contraption in the cell, called DNA polymerase, which fetches molecules of adenine, guanine, cytosine and thymine as required and slots them into place opposite the bases on the existing strand. It’s nature’s own form of high-fidelity photocopying. All life on earth depends on it, from the original oozy puddle three billion years ago through to 1998. But what if you were able to stop the process selectively? Stop it in a way that told you – up on the macro-level where enormous lumbering scientists move about – exactly which of the four bases had last been added to the chain? You take some single-stranded DNA and put it, along with a construction enzyme, into a medium where there’s plenty of A, G, T and C available. Only you’ve doctored the supply, so that among the ordinary bases there are some molecules that are a slightly different shape. If one of these gets randomly added to the growing strand, nothing will fit on the other side of it, making the chain terminate there; and the special ‘terminator’ versions of A, G, T and C are also tagged to make them detectable. (When Fred Sanger was first devising the method at the LMB in the 1970s, he made the terminator bases radioactive. In the ABI machines that automated his idea, the markers on the terminators became four different fluorescent dyes.) Of course, one little dot showing up where a chain ends won’t tell you anything. But you couldn’t put one piece of single-stranded DNA into the experiment anyway, since there’s no way to move individual bits of DNA precisely around the place. Instead, you’ve used DNA polymerase in its usual copying mode to run off a zillion copies of the single-stranded piece you’re interested in. There’s no way to predict where any particular one of the zillion copies will accidentally incorporate a terminator base and stop, but probability says that if you have enough copies of a relatively short piece you will end up with a mass of stopped strands that altogether incorporate a terminator at every single position in the sequence. Then you take your stopped strands and you gently inject them into one end of a flat leaf of acrylamide jelly with a mild electric current flowing through it. The current makes the DNA strands migrate along the gel, and with a beautiful predictability, the shortest strands are carried furthest. In fact, the distance the strands move is in exact proportion to their length. When you switch off the current, all of the zillions of pieces where the terminator base was incorporated at the very first available position are up at the far end together. Slightly short of them (but quite distinct, because the fragments halt together in a neat line) are all the pieces where the terminator happened to go in at the second position. And so on, right along the gel. Put the gel under a light whose wavelength makes the dyes in the terminators fluoresce, and you see that the neat line representing each position on the double helix is shining out in one of four colours. The bases in question stand only 0.34 billionths of a metre apart on the DNA molecule, but you have persuaded them to become visible a few millimetres apart in the gel. You couldn’t reach down to the DNA with a mechanical probe, but you just did it with a jointed, articulated probe of logic; and the logic works, the logic is secure. The knowledge it transmits from the realm of the ineffably tiny can be relied upon. Up at the end of the gel there, that’s a red line. Gimme a G! Then a blue one. Gimme a C! Red again. Gimme a G! Yellow. Gimme an A! Whaddaya got? GCGA! And that, in 1998, was how you sequenced the human genome.

  It was the straightforward middle part of the task, anyway. Before the actual sequencing came the selection, multiplication and preparation of the single-stranded samples. The ABI 377 machines with which the Sanger Centre was equipped could handle a piece of DNA about 400–500 bases long. Chromosome 1, allotted to the Sanger among other things in the public project’s grand share-out of responsibilities, contained about 280 million bases. It was the largest of the human chromosomes. Feeding it to the sequencing machines required far more than that it be snipped into the number of 500-base chunks that simple division suggests. It was cloned inside bacteria in 150,000-base stretches, randomly cut. The clones were mapped against the known features of the whole chromosome, creating a patchwork index of what went where. Craig Venter planned to miss out this stage and to trust entirely to pattern-finding algorithms to put the whole thing back together, but the public project believed that the mapping was essential as a quality control. Then the clones were randomly cut again, this time into sections small enough to sequence, but many times over, in different ways in the different clones, to achieve massive redundancy. The only way to know how to string together the 500-base sequences that were going to come out of the machines was to line up segments of the sequence that overlapped, and that meant creating enough different 500-base pieces to ensure overlaps that covered the entire chromosome end to end. Finally, all the new short pieces were re-cloned in bacteria, so that the sequencing machines would have the necessary volume of them.

  Altogether, it was a giant effort. And after the sequencing, another giant effort went into assembling the raw output. Huge amounts of computing power were directed at the overlapping 500-base fragments. At this point, when Venter was just throwing his hat into the ring, the Sanger aimed for 99.99 per cent accuracy. They called a sequence ‘finished’ when they were confident that they had reduced the error rate in it down to one base in 10,000, which took eight-fold coverage of the sequence, eight passes through the length of it, with adequate overlaps, gradually eliminating uncertainties and giving the search algorithms enough to go on. Not every pass produced unambiguous results. There were holes, there were persistent rough patches. Some regions of some human chromosomes put up an inscrutable resistance to being cloned in bacteria at all. The ‘finishers’ in the bio-informatics department were constantly sending back requests for specific portions of specific clones to be sent through again.

  The Sanger could call on plenty of gifted individual scientists to design these tasks and to supply judgement and interpretation: Sulston himself, of course, and Alan Coulson, who had been Fred Sanger’s assistant when terminator sequencing was discovered, and Richard Durbin, who wrote the database programme for the C. elegans genome and had adapted it for the analysis of the human data, and many others. But to lay out all of the tasks as industrial processes, flowing together unimpeded to form one high-volume production line, required a set-up quite different from the traditional organisation of a lab. John Sulston had begun sequencing the worm in 1990 with two machines and a handful of PhD students. It was the usual arrangement for a research project. But, he discovered, it didn’t scale. You couldn’t speed up reliably; you couldn’t go twice as fast by having twice as many PhD students, because the PhD students were not spending their time single-mindedly performing one activity at once, in units that would multiply. If you asked them to, they tended to become bored and resentful, for reasons Sulston understood very well. The gift economy of a lab is pre-industrial. It is a world where everybody is a generalist, pitching in to do a little bit of this and a little of that on the shared project, which everybody owns. You don’t do the same thing over and over. You do lots of things, so that you learn to do everything that people in your area currently know how to do. Some people are junior and some people are senior, but all the juniors can reasonably expect to become seniors. All the apprentices can expect to become masters. It was a good way to organise science, distributing responsibility and curiosity as widely as possible, and Sulston revered it. But it wasn’t adapted to getting a very large quantity of repetitive work done quickly. For that, he needed to break down the project into a stream of separate full-time operations, some skilled, some unskilled, some semi-skilled. He needed the division of labour.

  When the move to Hinxton was mooted in 1992, and the prospect of the Wellcome money started to loom on the horizon, Sulston realised that he had
better recruit somebody who would know how to execute the idea. Wellcome found him a finance man he got on with, Murray Cairns, formerly with the brewer Bass; but he also needed a scientific administrator, somebody who would be adept at the genomics and the management of the production process. Dr Jane Rogers is the director of sequencing at the Sanger Centre; in effect, general manager to Sulston’s chief executive. She was working at MRC headquarters in London, in 1992, commuting daily from Cambridge and rather regretting the sensitive-skin condition on her hands that had meant she had to move sideways out of active research. Sulston swooped on her one evening when she arrived at Cambridge station and drove her back to his house in Shelford, where he and Alan Coulson pressed glasses of alcohol upon her and tried to ask her the kind of questions they hoped would illuminate her suitability. ‘They plied me with sherry and asked if I knew how to turn an office building into a lab.’ ‘Did you get the impression that they knew how to do things like that themselves?’ I asked her when I interviewed her in 2002. ‘No, but they knew that they needed to know …’ She accepted the job offer. First, she sorted out their official grant application to Wellcome; then she set to work with Sulston creating and staffing a production line. ‘John had a vision. He saw that this was a repetitive task, so therefore you wanted people who could undertake repetitive tasks but would carry out the work well and also cared about carrying it out well.’ As Sulston puts it, ‘If you’ve got a job that would bore Bloke A, you hire Bloke B who is not at all bored by it and may actually find it quite amusing for a while … We decided that we would hire outside the box. We would hire people who were not qualified, in the conventional sense.’ It was a drastic break with standard lab practice. They advertised in the Cambridge Evening News, not in Nature. They hired lots of people with doctorates and lots of people with biology degrees, but lots more people who had just left school with minimal GCSEs or who were returning to the workforce after twenty years of child-rearing. They were looking for applicants who were careful, whether or not they had paper qualifications – who had steady hands and were willing to be trained to do extremely precise things many times in a row. Jane Rogers joked that Sulston seemed specially willing to sign up ex-barmaids. They arrived at Hinxton in 1992 as a group of fifteen. A year later, eighty people worked at the Sanger Centre. By 1998, it employed 500.

  The installation of the new model of organisation did not go entirely smoothly. Sulston willed the end, but he didn’t will the means to begin with. Not liking hierarchies himself, he tried to do without one at the Sanger Centre. ‘My approach to management was to say, well, I’ll do any job that needs doing that nobody else can do at the moment, but while I’m doing it we’ll try and find somebody else who can do it.’ His door was always open; he was on first-name terms with everyone. ‘The mistake I made was to think that was enough. Of course, it was ridiculous: when you get up to more than fifty people, two hundred, five hundred, they can’t all walk through the door at once, and they’re intimidated, and people need structures. I didn’t understand any of this. Remarks were made about piss-ups and breweries … I was mortified.’ Science’s system of informal authority didn’t scale either, it seemed. If production at the Sanger was going to travel along the lines of a complex flowchart of specialised tasks, there needed to be equally clear lines of responsibility. And alongside there had to be grievance procedures, and personnel reviews, and holiday entitlements, and gradually accruing pension rights. They put them all into place.

  If you had visited the Sanger Centre in 1998, on the eve of Craig Venter’s announcement, you would have seen a place that both looked like an academic lab and didn’t. Inside the new white buildings on the green lawns, office doors made of blond wood opened off quiet corridors. White coats hung in rows on pegs. Through glass walls, you could see the traditional apparatus of sinks and benches and fume cupboards in use. But the ceilings were higher and carried enormous ducts spiral-wound in insulating foil; the lab spaces were bigger than they would have been in a university department; and if you went into a room, instead of the harmonious disarray of ordinary research, you found everyone doing the same thing, again and again, to the sound of a cassette-radio in the corner playing that team’s particular choice of music. Fly me to the moon, crooned Frank Sinatra to those tending a robot programmed to pick speck-sized bacterial clones out of petri dishes. And the trumpets and saxes of the big band said goodbye to the picked clones as they passed out of the room on their trolley into the next room, where a different group loaded them, safe in the ninety-six separate divisions of a plastic box of growth medium, into a cabinet that would gently shake them for twenty-four hours, ensuring maximum reproduction of the precious lengths of human DNA, while Robbie Williams sang in the background. I’m loving angels instead … Meanwhile, ‘gel ladies’ in white overshoes trundled fresh supplies from the gel kitchen along the quiet corridors, and technical assistants in the humming chamber where the ABI 377s did their thing ceaselessly topped up the containers of reagent behind the machines’ access panels. Looks like quite a bit of congestion south of the city, burbled a Radio Cambridgeshire DJ – but heyyy, the sun’s out! Over and over, someone with a steady hand tipped a gel sheet on its end, pulled out a ninety-six-toothed sterile plastic comb which had made ninety-six little indentations in the setting acrylamide, and titrated in ninety-six tiny samples to await the electric current. Over and over: tip up, comb out, ninety-six little disposable pipettes. Tip up, comb out, ninety-six little pipettes. Over and over, to the point where Jane Rogers had to worry about these particular steady-handed individuals developing RSI, which had never been an occupational hazard before in the field of genetics research. It was a new world.

  Well, it was new to them. What they had reinvented, of course, was the Industrial Revolution. They had rediscovered the insight which Britain had given the world, for the first time in all human history, two hundred years and a fistful of decades earlier, when the Black Country began to blacken, and the spinning jenny started to spin, and the clumsy beams of the earliest steam engines started to nod up-down, up-down, up-down, in a rhythm that has never ceased since, though you would have looked for it in vain in 1998 in the cities where it first was heard. It was not primarily an insight about automation (although without it, you couldn’t harness automation’s power). It was this: if divided into specialised, repetitive components, human labour becomes, not just a little bit more productive, but exponentially more productive. Adam Smith explained the point in The Wealth of Nations with his famous example of a pin factory. Ten men worked in it, each of whom could maybe have turned out twenty pins a day if he had worked separately and manufactured each whole pin himself from start to finish. Together, on the other hand, by breaking down the pin-making process into tiny distinct tasks, they were able to manufacture not two hundred pins a day, not three hundred or five hundred or even a thousand, but 48,000 pins, day in, day out. ‘One man draws out the wire, another straights it, a third cuts it, a fourth points it, a fifth grinds it at the top for receiving the head; to make the head requires two or three distinct operations; to put it on is a peculiar business; to whiten it is another; it is even a trade by itself to put them into paper …’

  A trade by itself. Adam Smith used those words with conscious wonderment. He was witnessing something that had formerly been solid and considerable and lifelong become small and ad hoc and temporary. A trade had meant a craft, with a craft’s slow apprenticeship and its mysteries and its dignities. You became a candlemaker by sweeping the floor of your master’s workshop while you learned the secrets of tallow and beeswax. Then you saved for a shop of your own, and you walked in the Whitsun parade with the other candlemakers, wearing your best white apron. Now, suddenly, a trade was redefined as just a cycle of repetitive actions, like the putting of pins into a paper twist, which flashed momentarily into existence when a production process was broken down into a particular set of steps, and was liable to flash out of existence just as quickly, if the process was reconfigu
red. All over eighteenth-century Britain astonishingly specific occupations flourished for an instant, and then vanished again. In the 1790s, if you worked for Wedgwood in the Potteries, you could be a saggar-maker’s bottom-knocker. Delicate porcelain being fired in a bottle kiln had to be protected by rough clay cylinders called saggars. The saggars were thrown continuously on a potter’s wheel by a saggar-maker. The saggar-maker’s assistant, who walked along the drying rows of saggars removing the unwanted flat bit where the saggar had touched the wheel, was a saggar-maker’s bottom-knocker. In exactly the same way in the 1990s, if you worked at the Sanger Centre, it was briefly possible to be a full-time gel puddler. Not that they called it that – the art in question appeared and disappeared before having time to acquire a name. But the flowchart demanded it exist, and so it did. Acrylamide gel, recently reclassified by the Health and Safety Executive from neurotoxin to carcinogen, was not available ready-poured for the ABI machines. It had to be prepared by hand – gloved hand – on special tables with a built-in downdraught to carry away the vapours; and it turned out that there was a definite knack to achieving an even slick of gel with no bubbles between two sandwiching glass plates a millimetre apart. When ABI’s new capillary sequencers arrived at the Sanger, the task would vanish, because the ABI 3700 dispensed with gel. But as of May 1998, the Sanger’s whole operation depended on a steady supply of hand-poured gels. There was always work up at t’Sanger for a skilled puddler.