Wednesday, June 20, 2012

Giving a Scientific Talk to Your Department

The past four months have been a blur of activity for me. In addition to doing some pretty intense computational work, I finally received some of my first ever whole genome sequences from Complete Genomics, which I will get around to analyzing... eventually. Then, in what seemed like the blink of an eye, it was time to give my first ever open presentation to the Genetics Department at my university. Laying out your research (in progress) for all fellow students, postdocs and faculty to see (and critique) can be a bit daunting. In all honesty though, I think sharing your work with others is one of the most enjoyable parts of doing research. Not only is it great story telling, it's a great story because it's real.

Here's what I learned from my experience:

Before the talk
Start making slides early.
Because so much of my data involves number crunching and tables and graphs, it was getting really tedious sifting through folder and folder and list after list of files. So eventually I started keeping a running powerpoint of my work and using the comments box for descriptions. Maybe people already do this, but it didn't occur to me to start curating my work this way until I was swimming in data tables. It turned out to be a huge time saver because when it came time to put together my formal talk, I had already made figures for most of my main points.

Give a practice talk.
This serves the dual purpose of making sure you don't leave the presentation making to the last minute and also gives you an opportunity to get feedback on your analysis and presentation style. I gave a formal lab meeting one week before my talk, which helped other members of the lab get up to date on my work, and also gave me a chance to get some input regarding other analyses I might want to implement. Then the day before my presentation, I gave a slightly less formal presentation to my friends (some non-scientists, and some scientists outside of my department) to make sure I wasn't using too much human-specific jargon or glossing over essential background information. Both of these experiences were really useful, and I got some very good tips from both audiences. Lastly, giving a practice talk lets you ensure that you can cover all of your material in the allotted time.

The day of the talk
Get to the presentation venue early.
Our department talks are always in the same room, so this was less of an issue. However, it did allow me to set up on time and avoid the awkward five minutes when everyone watches the presenter fumble with the projector and display settings. Getting there early also let me realize that I had left my laser pointer and adaptor in lab, so I had enough time to run back and grab those.  Also, it just helps to get used to the room, so that you can focus on the data you're going to present, rather than distractions like the incessantly flickering fluorescent bulb near the door.

Once is not enough.

If a particular topic is central to understanding your research, it is not enough to mention it once. You need to constantly remind your audience about this fact. In my case, I created a navigation slide to give people some idea of where I was in my talk, and after covering each point, went back to a list of research questions and checked off the things I had already answered. Even so, there will still be people  who ask you to explain that topic yet again. This is just the reality of presenting to a room full of overspecialized individuals that dedicate only these thirty minutes of their time to your project. Remind people often what your research goal is and why it matters (despite what the best cynics might say, yes, your research matters).

Leave time for questions.
This can be hard if you have a lot of data to cover. But a deadline is a deadline. Think about the pieces of data that are most essential to telling your story. Other connecting bits may be interesting, but not essential. Leave these out. If it's really a missing link, your audience will ask you about it, and then you'll have the added bonus of being able to answer their question!

It's also a good idea to anticipate some of the questions you might be getting, especially regarding what you plan to do next in your experiments.

Be able to defend your answer.
There is nothing more embarrassing than when the presenter is obviously thrown by a question and rambles their way through it. The audience would gratefully prefer a succinct "That's a great question. I'm not familiar with any data that explains that." Alternatively, there's the even shorter, "I don't know, but I'll look into it." Or the even shorter, "I don't know." Gasp.

After the talk
Be gracious.
Be gracious to the audience for their time and attention. Even if half the audience is there for the free coffee and bagels, everyone in the room should have at least some marginal interest in science. So give them a scientific presentation. Make it visually appealing. Speak clearly and look at your audience, not your slides. Also, be gracious to the people brave enough to ask questions. Lastly, make sure you have an acknowledgements slide. I was really humbled when I saw just how many people put in time and effort along with me to answer "my" research question. Thank your advisor, your lab, your collaborators, your funding source, your audience, and above all else, the patients you've never met. Without their consent, you'd have nothing.

Treat yourself.
It's no small task condensing a year's worth of research into thirty minutes. Once your laboratory-house is in order, and your PI gives you their two thumbs-up,  treat yourself...
...With fancy breads and cheeses

...With fancy herbed potatoes

...With sage butter.

Treat yourself.

...then don't treat yourself again until you publish something (I'll let you know how that goes).

Thursday, February 23, 2012

A Case for Federal Funding in Biomedical Research

For the current fiscal year, the US government is proposing a flat NIH budget. When one considers the rate of inflation and the ensuing devaluation of our nation's currency, tomorrow's dollar will not stretch as far as yesterday's. What does this mean for today's research scientist? It means having to sacrifice manpower and equipment needed for biomedical research. It means fewer funds to train future generations of scientists. It means less funds available to fully staff research labs. It means my PI might be able to afford the reagents for my experiment, but not validation. It means I might make a discovery, but I'll be too broke to attend the professional meeting where I can share the results of said experiment with the broader community where said work might have an actual impact.

While I am as irritated with Congress as the next American about our nation's fiscal irresponsibility in recent years, biomedical research is one area that we should be fostering, not relinquishing. Federally funded research pioneered the development of affordable sequencing technologies, aided in the completion of the human genome project, and spurred the identification of therapeutic drug targets for a plethora of human diseases. Outside of human genetics, federally funded research has been instrumental in providing an environment that fosters discoveries that have been essential for understanding the basis of diseases like cancer and diabetes. Computational biology research has helped model the spread of disease, solve the structure of things like the ebola virus, and simulate population expansions during the course of human history.

Some biomedical research is also funded by the private sector, and it should be, but that funding is market-driven and goes after solving diseases that affect the largest number of people and hence can provide the largest returns on the investment. Yet there are a great number of individually less common diseases that together still affect very large numbers of people, and the private sector will not provide the money or manpower to understand and treat those diseases. As someone who studies Crohn's Disease, a disease that affects 1 in 100,000 people, that number might not seem like a lot. But that's over a million Americans with a disease that is not yet fully understood, but where progress has been made and where there exists the potential for it to be solved.

If you believe that biomedical research is a worthy endeavor of the federal government, please sign this petition in order to convey to our Congress how off-base they are on this issue. I know it's a long shot, but I prefer in this case to say something rather than nothing. You can sign the petition here:

"No endeavor that is worthwhile is simple in prospect; if it is right, it will be simple in retrospect."
-Edward Teller, Hungarian-born American nuclear physicist, b. 1908

Thursday, February 9, 2012

Bringing Science out of Yale and into New Haven...

When I'm not phasing genotype data, part of my life outside the lab involves organizing talks with the Yale Science Diplomats (follow their blog here). Last year we started a Science in the News Series in order to explain some of the science behind controversial headlines, like this one. Science Diplomats are a fantastic group of grad students who, come hell or high water, are really committed to sharing a love and understanding of science with the public. Last year I participated in a talk with my friends Yixiao and Eric on Personalized Medicine focusing on genomes and disease, pharmacogenetics, and stem cells, respectively. People of all ages came out to the Leitner Observatory to learn about what a genome is, what it tells us (and doesn't tell us) and how we can use (and are using) that information to predict, diagnose, and treat disease.

One of the best things that came out of that event was having educators from local public schools in the audience ask us if we'd be willing to bring the talks into the classroom. Keerthi (who keeps a nifty blog of her own you can read here), our outreach coordinator, organized everything for us. After lots of back and forth between a number of schools and finding a time for our talk that fit into New Haven's rigidly packed curriculum, we were able to visit a local high school and talk about awesome science.

This particular high school is something of a rarity. Called the High School in the Community, "HSC" was started by teachers who were frustrated with the state of public education and decided to take teaching into the streets. They were so effective at engaging students that eventually administrators woke up and offered them some physical space (an old factory on Water Street) that has been repurposed into classrooms. Some of these rooms may have poor lighting and no running water, but that does not take away from the amazing job the teachers are doing to turn that space into a full-fledged science explosion. I was blown away by the job biology teacher Stephen Zepecki has done to make sure his classroom is teeming with life--only fitting for a biology class. The walls are lined by giant aquarium after aquarium (maintained by the students) filled with all sorts of marine and reptilian life, and there are some veritable flora and fauna in the center of that classroom (it helps that there's a sky light).

^The most biological biology classroom

There is really something to be said about a person who teaches introductory high school biology and is obviously liked by his students. Seeing the rapport this class had with their teacher made it a little intimidating to get up there and give an engaging talk on molecular biology! Yet even from the sidelines, Mr. Zepecki kept the class tuned in. For a school where 40% of the student body has special learning needs ranging from dyslexia to ADHD, you wouldn't know it from the Herculean attention these kids maintained. They're also probably the most respectful audience of high school students I've ever witnessed. It's a testament to both their teacher and their own desire to learn. We're really grateful he let us into his classroom today.

^Steve Zepecki, his class, and Eric

I won't go into the details of the talks because they're viewable on the Science in the News YouTube Channel, but I will say it was a learning experience for us presenters as well as the audience. Sometimes as scientists we fall into the perilous trap of adopting jargon and forgetting how to distill information into its most fundamental and interpretable pieces, and important messages get lost that way. Having graduate student-researchers talk science with high schoolers helps both parties; the high schoolers are exposed to something new, and researchers learn to deliver information in a way that actually conveys what they are trying to say.

A few things to remember for our next classroom visit are that even though most people have heard of DNA, cancer, and stem cells, those concepts haven't necessarily been defined to this age group. Students aren't particularly forthcoming about the limits of their current knowledge, so it's our job to ask them where they're at and if they're following us. Also, repetition is key. In retrospect, "single nucleotide polymorphism" is something that I should try to say five-times-fast before moving on to what it's used for. The other thing I'm taking away from this visit is that the Socratic method is really invaluable. There is nothing like having someone arrive at an answer to their own question when they're given the opportunity to think through it critically. Yixiao in particular was very good at this today. The class really woke up when they figured out they'd be accountable for answering their own questions and testing the assumptions on which their answers rested.

^Yixiao being awesome.

I'm really impressed with this class. In 90 minutes, we covered the human genome, single nucleotide polymorphisms, single gene disorders, polygenic disorders, risk prediction, direct-to-consumer genetic testing, the Genetic Information Non-Discrimination Act, pharmacogenetics, biopsies, receptors, drug targets, survival curves, cell culture, genetic reprogramming, embryonic stem cells, adult stem cells, and induced pluripotent stem cells. And they took notes. I'm not advocating a crash course like this be the norm, but if all we have is 90 minutes, make each minute count. We shouldn't be so scared to talk science with our kids. They can handle it. They had some pretty insightful comments, too.

My favorite part of science isn't the benchwork or the analysis. It's sharing ideas with other people. It's making others aware of something that only seconds before was inconceivable to them. When you make someone's jaw drop, you know you've done something worthwhile.

Wednesday, February 8, 2012

Direct to Consumer Genetic Testing: Perceptions and Predictions

As many geneticists have pointed out repeatedly over the years, the cost of DNA sequencing is dropping at an exponential rate. Human genomic information will soon be utilized in the personalized medicine age. But for people willing to bank on one percent of their genome instead of all three billion base pairs, that age is already here.

In 2008, TIME magazine called the launch of direct-to-consumer (DTC) genetic testing company 23andMe the invention of the year. When it initially launched, it promised to predict your health risk for over a hundred complex traits, two dozen single gene disorders, a handful of pharmacogenetic profiles, and genetic ancestry using half a million markers for around $500. Today, 23andMe offers the same analysis for $99. And that appears to be the magic number. Thousands of people have since opted to have an Oragene kit shipped to their door, spit about one teaspoon of saliva into a purifier tube, mail it off, and wait two to six weeks for their genetic fortunes to be read.

While other DTC genetic testing companies may have come first, they have not been the most appealing. Noted one recent 23andMe participant, “When I look at other [DTC genetic testing] websites, it looks like they’re operating out of a basement.” Indeed, other genotyping companies with essentially the same technologies and services as 23andMe have pretty much ended up in the gutter. DECODE, the parent company of DTC genetic testing company DECODEme, went bankrupt last year, and another company, Pathway Genomics, was significantly curtailed when the FDA halted the planned sale of its DTC genetic tests at Walgreens nationwide. Part of 23andMe’s success in the DTC genetic testing arena can be attributed to their creative marketing campaign, their user-friendly website, accessible tutorials on genes and inheritance, and a little brand recognition and star power that went a long way (23andMe cofounder Anne Wojcicki is a Yale alum and the wife of Google founder Sergey Brin, and both Warren Buffet and Jimmy Buffet were enthusiastic 23andMe participants).

While other companies sought to define themselves as experts dispensing vital health information first and foremost, 23andMe simply advocated that DNA was fun, was something you had a right to know about, and perhaps most tantalizing in our social networking age, was something that you could share with others. And in an age of people with shrinking attention spans, they promised to deliver this shareable information relatively quickly. In an age where painfully inefficient bureaucracy awaits us at every turn, part of 23andMe’s appeal is the pace at which data is collected, analyzed, and shared. In a piece for Wired magazine last year, journalists asked an NIH official how long it would take the NIH genetics core to identify the genetic basis of a complex trait like Parkinson’s disease using 500,000 genetic markers and predict risk for disease. The answer was telling. What the NIH required six years to accomplish, 23andMe completed in eight months, and arrived at the same results.

Granted, some of the reported traits are of questionable use—who cares if you can smell asparagus metabolites in your pee?—and some of the predictions are downright wrong. I experienced this first-hand when one of my results informed me that I have curly hair (my hair is so pin-straight that even the cruelest of heating elements have failed to help it defy gravity). Thankfully this is a mistake regarding something trivial like hair texture, but if the site is providing inaccurate predictions about physical traits, how many of its health predictions for alcoholism, various cancers, and bipolar disorder, among others, are equally erroneous or at best, premature?

The risk for genetic misinformation at this early stage is quite high for a number of reasons: the discovery sample might be of an ancestry that differs from that of the participants; not all contributing loci are known; there could be unidentified gene-environment interactions, etc. Glossing over this fact gives people a false sense of confidence. Says one user of DTC genetic testing company Navigenics: "We hear a lot of different – and sometimes conflicting – opinions about how to take care of our health. I’m very excited about receiving only the most relevant information to me, based on my DNA." This is not an entirely accurate statement, however. While it is true that particular interventions such as certain diets or exercise routines might only be beneficial to a subset of people with a given set of genetic determinants, the risk predictions for diseases reported by these DTC genetic testing companies are still population-based, and conflicting genetic reports occur just as often as conflicting intervention reports. And unless those genetic risk predictions consider all known markers, the risk prediction you receive is very likely wrong. Take for example Francis Collins’ own experiences with DTC genetic testing. Collins, most famous for identifying the genetic cause of cystic fibrosis, for pioneering the human genome project, and most recently for heading the NIH, clandestinely submitted his own DNA for testing to three different DTC genetic testing companies. His predicted risk for prostate cancer varied dramatically from one test to the next, ranging from 23andMe’s “not at risk” to Navigenics’ “elevated risk of 40%.” How was this possible? While these three companies had roughly similar coverage of the genome, the coverage of known prostate cancer risk loci varied substantially among the providers, ranging from 3 SNPs in 23andMe’s test to 18 markers for Navigenics. Given that three similar products resulted in such discrepant predictions, it seems obvious that the release of these services for complex disease risk prediction was premature.

Daniel Goldhill, a third year in Yale’s EEB program, says he would only consider purchasing a 23andMe product for its carrier testing of single gene disorders, since “there are a lot of known disorders associated with Ashkenazi Jews.” He is less interested in the risk prediction for complex traits, however: “Why do I want to find out now if I’m at risk for some terrible condition when I’m fifty? Plus it depends how accurate [the prediction] is. Why look only at sequence data? What about gene expression? What about microRNAs? Epigenetics?” Laurel Hochstetler, a first year MCGD student and recent 23andMe participant, echoes that sentiment: results “should be taken with a grain of salt. They’re using such a small number of markers and supporting papers. There’s so much more to it...papers aren’t infallible, even if they are peer-reviewed”.

23andMe might insist that they are merely providing information with no intention of dispensing medical advice, but nevertheless, the majority of the information contained in their reports is genetic risk prediction for nearly one hundred complex diseases. As Richard Lifton once put it, this basically amounts to “practicing medicine without a license.” While there are many non-medical reasons for pursuing DTC genetic testing (adopted individuals who lack family history but are curious about their ancestry or want to find relatives, for example), there is a consistent theme of taking control of your health information. Despite calling it a “terrible idea”, Lifton conceded “what they’ve demonstrated is how many people are eager to do this.” Hochstetler agrees. “I just wanted to compare. Part of me loves to see all the combinations people have.” Despite have incomplete medical information, 23andMe has managed to get its foot in the DTC genetic testing door by catering to individuals’ curiosity, which will serve it well when it plans to use its swath of genetic data for its own participant-driven research endeavor, “23andWe.”

One interesting question will be how 23andMe will handle the shift from SNP genotyping arrays to whole-genome sequence data. Says one cell biology student, “I think it would lead to the world being like GATTACA, designing our own children. But if it takes more than 5 minutes, I’m not doing it. Maybe if I was bored on a rainy day or something.”

This entry is a re-post of an article I wrote for B, originally printed here.