Mind the Gap

As it’s Friday pretty much everywhere at the moment, I thought I’d share a little revelation I had this morning.

Every time I work down in the yeast lab, I get the nagging, uneasy sensation that someone is watching me. It doesn’t make any sense, because the lab is bright, cheerful, infused with music and laughter and banter – not even remotely sinister. But even when I’m alone, I feel those cool, dispassionate eyes boring into the back of my neck, prickling at the tiny hairs on my skin.

This morning, I finally worked it out: there is an undercover Dalek in the room, just waiting for the best moment to make his move. Can you spot him in this picture?

Scientists are trained to read the scientific literature with skepticism. Forever question, we are told, the truth of various assertions put to us no matter how eloquent or famous the writer or prestigious the journal in which that writing appears. In thinking about it more closely, though, I was surprised to recall that my first lesson in skeptical reading did not actually derive from my scientific training, but from an undergraduate course in Cultural Anthropology.

I got mixed up with Cultural Anthropology because Oberlin College bestows a liberal arts degree, so its humanities students must take a number of science courses and its aspiring scientists, classes in the arts and humanities. Although my classmates groaned about this, I was in heaven, gobbling up whatever I could – Ancient Greek, ethnomusicology, Spanish poetry – I even got credits for joining the women’s Ultimate Frisbee team and learning how to play the steel drum, the Gambian kora and several teeth-grittingly atonal gong-like instruments from the Javanese gamelon tradition. But that’s another story.

The formative lesson in question was conducted by a wily professor who assigned Margaret Mead’s Coming of Age in Samoa (1928). For those of you not familiar with this work, it’s chock full of teenage sex and its central theory was that Samoan adolescence wasn’t stormy precisely because nothing, including sex, was taboo. After giving us a chance to marvel at Mead’s achievements, he then set the text Margaret Mead and Samoa: The Making and Unmaking of an Anthropological Myth by Derek Freeman, which basically claimed that Mead’s findings were biased and artifactual. So which anthropologist, we asked the Prof anxiously, was right? He shrugged and said that the truth would probably never be known, as numerous scholars had found problems in both Mead’s and Freeman’s methods, and the culture in question was long gone.

As a callow girl of nineteen, I found this whole affair rather shocking. How could something in a book be wrong?

Of course I soon learned that many printed things are wrong – and this was years before the internet taught everyone to take everything written publicly with a grain of salt. Still, I had a healthy respect for what I read in scientific papers, taking everything at face value unless I had a valid reason to believe otherwise.

But that’s all over now. I can’t decide if it’s a blessing or a curse, but suddenly I’m having problems believing a single word I read in the scientific literature. It all started a few months ago. With only a year and a half left of my fellowship, I’ve been thinking about the long game, prioritizing my activities, clearing the decks to embark on a tangential line of research that – with good luck and a following wind – I might be able to export from this lab to start my own. With a new project entails reading up in a new field – always a painful prospect when you don’t know the background or any of the lingo.

And there I was, lost in a dense review article paragraph about a pathway with dozens of proteins, with laboratory A saying protein W does P, and laboratory B claiming no, protein W does process Q, and laboratory C begging to differ, that W might bind the main effector of process P but under different conditions, W probably got co-opted to perform process R, while simultaneously initiating a positive feedback loop bolstering process P and Q. And I was clobbered: clobbered by the incredible complexity of this seemingly simple biological phenomenon; clobbered by the number of papers, many contradictory, chipping away at the problem, each revealing yet more complexity; clobbered by how bitty and insignificant each finding appeared against the larger messy advancing-and-receding scrum of raw observation, supposition, conjecture, and dissent. OK, laboratory A might have published it in Nature or Cell, but suddenly, it just seemed – in the grand scheme of things – unbelievable.

A little bit of skepticism is a good thing, but I feel like I’ve been infected with some terrible virus.

Is it terminal?

If science is a narrative, then sometimes you have to read between the lines.

Consider, as case in point, the humble Acknowledgements Section of your average peer-reviewed journal article. I have always been fascinated by this little afterthought of a missive, consigned to the end after all the “important stuff” and, unlike the rest of the text, probably dashed off at the last possible second. On the surface, it’s a rather relaxing and positive little place: the authors thank X, Y and Z for providing key reagents, and members of their laboratory for helpful comments, and colleagues for proofing the manuscript, and funding bodies for providing the dosh to make it all possible. Here is where a bit of human warmth is finally allowed to permeate the barren tundra of scholarly wisdom; you can picture sepia-tinted senior common rooms glowing with amiability and good sportsmanship, tea being poured and biscuits being passed around (and no one ever helping themselves to more than their fair share).

But alas, interactions amongst scientists aren’t always so collegial. In fact, one’s peers can be downright petty, obstructive and sometimes even malign. In consigning only goodness to the Acknowledgements section, we’re missing out on a brilliant opportunity to tell the story of our work in all of its entirely. In this spirit, I hereby propose introducing a new section to the standard scientific manuscript: the Disacknowledgements Section. This would honor all the people who obstructed the research; the cast of characters – shall we refer to them as anti-authors – without whom the work would have proceeded much more briskly and less painfully.

Below, I highlight a few personal favourites from my own scientific career. (Although I’ve tweaked details here and there to obscure identities, the fundamentals of the examples are all true.) Feel free to chime in with your own – I can assure you it’s quite cathartic.

• The authors wish to disacknowledge Dr X, who sat on the graduate student grant committee charged with judging the first author’s third-year fellowship (proposing to hand-sequence thousands of viral genotypes to search for patterns in mutation during disease progression), and who was instrumental in rejecting it because he thought there was “no way any student could sequence that much DNA in one PhD stint”. Four years, a megabase and six first-author papers later, she begs to differ. (To be repeated in the Disacknowledgements Section of all six papers, with gusto.)
• The authors further wish to disacknowledge Professor Y for insisting that not only she, but also her post-doc, be authors on the paper for providing a small bit of patient control tissue. Please note that this study also used human tissue provided by seventeen other individuals, none of whom thought that their effort merited authorship – and indeed, they would have been embarrassed to be honoured thusly. (See Acknowledgements for these lovely individuals.) Note that Professor Y had professed herself completely uninterested in the project, and indeed strongly against it in principle, until she found out what surprising data came out of it and what a good journal it was going to be submitted to; even after muscling in on the act, she never did read the manuscript or had any idea what it was about – and neither did her post-doc.
• The authors also want to disacknowledge minor co-author Dr Q who, not being a native speaker, nevertheless read the entire manuscript and made only one comment: that her collaborative data would be pulled from the paper if the first author (a native English speaker) did not change the word “material” to “tissue” when referring to the samples used in that assay.
• The authors would like to disacknowledge the co-last author Professor B, who threatened to pull his data entirely from the manuscript if the third author, his PhD student, was not promoted to second author. We would like to note that this bogus second author performed only one Western blot in isolation during a rotation (see Figure 3F for her contribution) and had little idea what the main thrust of the manuscript was about. Meanwhile the demoted third author had been working on the project for three years, was responsible for two of the key findings, but was (to quote Professor B) “just a technician.” Readers might also be interested to know that in the country in which this paper’s lab resides, second authors are allowed to include such papers as chapters in their PhD thesis. Coincidence? We don’t think so. (The first author is also a little cheesed off at the last author for caving in to this outrageous demand “for political reasons”, but after throwing a few pipettors across the room and calming down a bit, she did understand that, in not having tenure, he had little choice).
• Finally, the authors would like to disacknowledge Bigshot Professor P, who sent his post-doc halfway across the world solely to spy on the first-author’s poster and pump her for a few helpful details. Once the first-author cottoned on, she ended up having to publish the story half-finished in a minor journal to avoid getting fully scooped by P’s lab, which was about ten times the size of her own. (This is a lovely, nail-biting story of intrigue and sculduggery in its own right, which might end up one day soon in the first author’s blog as a separate post – or possibly in one of her novels.)

There’s a new kid on the block – in fact, it’s housed in our laboratory, despite being a communal piece of apparatus. Rumors of its arrival buzzed in the institute’s corridors for weeks, and when the inevitable training sessions were organized (by a post-doc whose eyes glowed with evangelical fervor), I avoided the opportunity to enrich myself. In fact, the momentous appearance filled me with a diffuse sense of melancholy.

Why? Because I knew if I tried it, I’d like it; and if I liked it, I’d use it. And if I used it, I might never frequent a darkroom again in my scientific career.

I’ve talked before about new tech and kits, how they can make lab life easier and experiments more reproducible. And if I despise the old-fashioned protocol being supplanted – phenol extractions, say, or old-school cuvette spectrophotometers – I will embrace their modern replacement with enthusiasm. And yet – I’ve also talked about how much I love developing films in darkrooms, and how this ritual has being inextricably linked with all of the memorable moments in my scientific career. So for me, computer imaging of chemiluminescent protein blots was always going to be a tough proposition.

Dear reader, I gave it a go – and it’s amazing. I probably won’t look back. Yes, I’m sad, but on the other hand I’ve got quick, clean results, all tidily scanned and ready for publication. There are a few drawbacks: one person can only develop one blot at a time, so there are bound to be big queues as the technique catches on. Also, it’s a serious pain trying to line up your non-superimposable rainbow markers in greyscale. On balance, however, defection appears to be in order.

So it’s farewell to safelight, Kodak film, X-ray developer, the rumble of the barrel door, the trickling of water, the pungent solutions and, above all, those quirky anonymous interactions with colleagues unseen. I’ve come out into the light.

They’re pale and creamy, scattered across the agar like diseased lesions. Something about the three-dimensionality of the colonies – vaguely dome-shaped instead of flat – rings some sort of primordial alarm bell deep in my hindbrain. Although I know it’s irrational, I can’t control the little shiver of revulsion each time I stab one of the glistening disks with a sterile toothpick and spread it over the next pie-shaped zone of the fresh plate. Even the way these yeast people have taught me to streak seems inherently wrong: my Microbiology tutors at university would be horrified by the laissez-faire way I am now scribbling onto the surface of the agar, with no system or intent to isolate single colonies.

Of course I am thrilled that all of my transformations yielded so many resistant colonies, but to this biomedical researcher, yeast are only ever bad news: they infest your tissue culture cells, grow inappropriately on your bacterial plates and cause all manner of nasty infections in patients.

So, after months of sitting in front of a computer, I’m finally back at the bench. Just today I hit the shiny ‘submit’ button that should consign my further revised manuscript to its final acceptance (seeing as how the editor had only asked for minor text changes); and after a visit from my collaborator last week, it was clear I couldn’t proceed further on data analysis until he’d performed a few more bioinformatical dark arts with our dataset. All this meant that I was free to escape to something more interesting in the interim.

So escape I did, to my lovely new collaboration with the local Schizosaccharomyces pombe guys down on the first floor. Even after a few days with this crew, I am starting to suspect that fission yeast people are inherently cooler than your average biologist. For starters, they listen to music in blatant disregard to institute rules – not loud enough to trouble anyone, but just loud enough to give off a rebellious vibe. And of course, the entire place smells faintly of beer.

It’s a brave new world: I’ve just made a series of mutants of the pombe homologue of one of the most promising novel hits from my cell shape screen. In one week, you can knock out your gene; you can replace it with a tagged version of your gene under its endogenous promoter; and you can put in a version of your gene under an inducible promoter, all down to the magic of recombination. We made monster recombination cassettes using 100-mer PCR primers and then just zapped them into the bugs – the pombe did the rest. I estimate that it’s taken me ten days to achieve something that would have taken two to three years in a mouse.

It’s already exciting: I can see that my deletion mutant transformant colonies are noticeably smaller than the wild-type, so there might be some sort of effect on growth. In a few days I’ll be able to peer down a microscope and visualize the tagged protein to see where it is localized in the cell, and I’ll be able to look at the knockout and overexpressing cells in more detail to see if I can discern the nature of the defect. Of course once I have a clue about pathways, there is an entire genomic library of yeast strains that I can mate my new strains with, to look for genetic interaction all in a matter of a few days.

And the best bit? By convention, I get to name these new yeast strains after my own initials, and they will pass forever into the global community’s strain collection. JLR-1, -2, -3 and -4 may not be as exciting or poetic as a Drosophila gene name, but it’s a hell of a lot more personal.

I was gripped by a sense of mounting panic as I stared at the diagram in front of me, shoulders tensed and pen clenched in my right hand. I’d been bashing away at the thought experiment for well over fifteen minutes and, somewhere beneath the immediate problem was the wider realization of just how bad this could be for my overall project if I failed. The chatter of the office had receded into a blur of background noise: it was just me and the thought experiment, battling it out for world domination.

I’ve been finalizing the annotations on my final list of gene hits from my big screen for the past few weeks, transferring the features of hundreds of images into a yes/no style spreadsheet. Although I had long since scored individual pictures blindly, I was now in the phase of breaking the code and examining gene behavior in the round. I first realized that something was wrong when I noticed that one of the negative controls on a particular plate had a very strong, distinctive phenotype, whereas on the same plate, a positive control gene of known function failed to make the expected morphological impact. My colleague and I had performed the screen in duplicate and the anomalous results were perfectly replicated, for various parameters.

So this told us something: whatever had gone wrong had gone wrong at the very beginning. The original RNAi library had consisted of 32 individual 96-well (“mother”) plates, which a robot had arrayed into eight 384-well “screening plates”, over and over until we had enough replicates. Mother plates 1-4 were arrayed into screening plate 1, mother plates 5-8 were arrayed into screen plates 2, and so on. Now, when a robot arrays four plates into one, it doesn’t transfer each 96-well plate en masse into one quadrant of the larger plate, like this:

Instead, it arrays the same site on each four plates into a quadrant, either clockwise or anti-clockwise (the former being the case for my screen):

At first glance, from the pattern of some of the controls, it looked as if the problematic screening plate had simply been placed on the robot rotated 180 degrees, which would give a pattern like this:

But various clues, especially from internal regions, suggested that the problem was much more insidious: than one, some or all of the 96-well mother plates had been placed on the robot rotated 180 degrees. If just one plate had been misplaced, you’d get a pattern like this:

But if all four had been misplaced, it was much more insidious, especially as many of our controls end up being arrayed into quadrants by virtue of their original plating, meaning that if there was an internal shuffle within the quadrant, you’d never know:

So there I was, sitting at my desk, trying to work backwards from the phenotype on the problematic screen plate to the original source destinations to see if I could figure out how they had been swapped. Of course cell phenotypes are not as clear as pretty colors, so it was really difficult. But that wasn’t my biggest problem. My biggest problem turned out to be my poor spatial awareness. Women, we are told, are not very good at this sort of task. I don’t know how it is for other women, but when I’m at an aerobics or karate class and the teacher is facing the audience, I struggle to mimic her movements in mirror image. When I’m lost, I brazenly turn maps upside-down so I can more easily work out whether to turn left or right. So the current exercise was like my own special room in Hell. I briefly contemplated sweet-talking one of our computational biologists into writing me a program mapping out the fate of the colored tiles under all the different flip permutations – I mean, compared to modelling cancer or gastrulation, how hard could it be? I’m sure some people could have done it in their heads, but after those fruitless fifteen minutes, I decided to surrender gracefully to my own limitations, and found my very own discipline: paperinformatics. In other words, I got out my colored pens, got to work on a lot of sketches and soon had the problem sorted.
The verdict? All four mother plates were one-eightied, so all I had to do to repair my annotations was to swap each well designation diagonally in each quadrant. After doing so, suddenly all the weird phenotypes and anomalies fell magically into place, and I was back on track.
The moral? No matter how tired, rushed or hungry you are when using a liquid-handling robot to plate out a major screen, always, always, always triple-check the orientations, not only of your mother plates, but of your destination plates. If my controls hadn’t been so conspicuous, I might never have known, and would have ended up with a plate full of false results. I might even suggest that it might be a good idea for plate manufacturers to start making the A1 corner much more conspicuous than most brands currently do (a raised black letter on a black background, or a subtly blunt corner contour). How about a bright color for the spatially-impaired?
I’m sure the focus groups will tell you to make it pink.

I may work with some of science’s finest minds, but there comes a point when you have to bow to the inevitable.

Leaving notes in the lab simply does not work. Those who notice it will just think you’re a bit prickly or premenstrual, and those who don’t – probably your intended audience in most cases – won’t pay heed anyway. So all in all, there is little point in the exercise, and I’ve decided to officially surrender. It wastes paper, and in all the fuss of finding your pen (nicked by a labmate, no doubt), it wastes time too.

Now if you’ll excuse me, I have someone else’s yellow bag to cart off to the lift disposal area. Why does my heart feel so light?

It’s crunch time.

I’ve cleared my desk for the final assault. No more distractions, no more experiments: I’ve even thrown away all my cell cultures so that they can’t tempt me with their gossamer lamellipodial embraces, with their scandalously promiscuous ability to take up inhibitory RNAs. Nine months after agreeing to “just quickly make three constructs” for someone else’s project, I have submitted what is hopefully the final revision of a paper that more or less ended up entirely in my lap and went three rounds with various journals. Nine months: rather symbolic, isn’t it? Those three constructs turned into several figures once the first author moved on, and my name moved up on the author list, and eventually I was responsible for all the myriad minutiae that birthing multiple versions of a paper requires.

Was it worth it for a second authorship? I honestly have no idea. But it felt good, after having addressed the lion’s share of the major revisions, to press that beautiful blue button marked “submit”.

Did you press the button yet? my boss emailed me, midway through the afternoon.

Not only did I press the button, I emailed back, but I stuck a wooden stake in it for good measure.

My own projects have lain fallow for several months now as I wrangled this paper to what I hope will be its final resting place. But now, things are going to change around here. No more Dr Nice Girl. Allow me to introduce you to my new Universe:

Until further notice, I will be eating, drinking and sleeping with my final gene list. We’ve done the screen in duplicate now, and my colleague and I have scored all the data visually, and now I’m in the process of arriving at a Venn decision about the entire dataset: does this gene affect cell morphology robustly, reproducibly and with the majority of four non-overlapping inhibitory reagents, or does it not? And if so, each visual phenotype must be described by a controlled vocabulary of over fifty descriptions.

There is a shocking amount of noise in our system. We knock down every gene with four different siRNA oligos, but many of these oligos have different effects even though they are meant to target the same gene – although thankfully, this is entirely reproducible. I am tracking these problem cases down now, eliminating the genes whose oligos just can’t agree on an overarching morphological effect. I am being ruthless, and it feels good. My longlist was over 300 genes, and the boss won’t be happy with more than 100. So out they go, left and right. I don’t care how beautiful the phenotypes are: star-shaped, sharply triangular, wispy and neuronal: if the oligos don’t agree, it’s sayonara.

I can’t stare at these images and enter their data into spreadsheets all day, though: I’d go mad. And it’s not just mentally difficult: my neck and head eventually start to ache, and the chronic repetitive strain condition in both of my hands is nearing critical. So for light relief, I get ahead on much-loathed chores, like tidying the tissue culture room or consolidating and alphabetizing the lab’s chaotic individual stashes of restriction enzymes. When people ask for my help I leap to their aid desperately: yes, I’d love to watch you develop your ECL on the new ImageQuant machine, or to share my recipe for the perfect SDS-PAGE running buffer! If you’re too busy, I’ll even make it for you!

I’d forgotten how lovely and cold Tris/Glycine running buffer goes when you prepare it, the reaction sucking the heat out of water until the entire bottle is frigid. At times like this, when research gets far too abstract, it’s a relief to experience the mysterious laws of nature up close and personal, reminding me of why I’m here in the first place.