Académique Documents
Professionnel Documents
Culture Documents
SPOTLIGHT ON BIOINFORMATICS
THERE ARE two paths to years ago. As speed has gone up and
DNANEXUS
Its important to careers in bioinformatics, both costs have come down, demands on
of which require learning a new bioinformaticians have grown. They
enjoy your job and be language. Computer scientists need to be comfortable with mining
must become fluent in the life much larger data sets, and looking
motivated: the best science terminology of genetics, for relationships between them.
genomics and cellular biology. Asimenos had to first get
ideas come whilst Im Biologists must pick up skills in comfortable with the language
data analysis, including statistics, of biology before he could dive
in bed, or walking my logic and programming. When into the data. He remembers
the field was developing, fledgling hearing terms like 3 Prime
dog, or having a coffee bioinformaticians often taught and downstream and thinking
themselves. Now, more institutions What do these things even
with colleagues. are offering formal training, and mean? How do I find out?
the field is maturing rapidly. He gave himself a crash course
Federico Abascal The skill set needed by a by reading textbooks, going to
bioinformatician continues to conferences and hanging out with George Asimenos
evolve. In the early days of the biologists. I had to overcome the
human genome project, it was vocabulary barrier, he says. During
sufficient for scientists to find his undergrad, Asimenos had taken
homologous genes of one organism courses in statistics, engineering in statistical genetics analysis at
in the genome of another. Now, and computer science. Acquiring deCode Genetics in Iceland. He
bioinformaticians routinely those skills earlier gave him time focused on how statistics can be
compare multiple genomes, during grad school to bone up used to find areas in the human
analyse regions that dont code on his biology, literally, with an genome that contribute to increased
for DNA, and incorporate a anatomy class that included human risk of disease. Now, as the director
host of proteomic information dissection; and figuratively, with of the Centre for Computational
in their analysis. Both the type stints in molecular biology wet labs. Biology, University of Birmingham,
and amount of information Still, he had some fraught he has been tasked with teaching
continues to expand, as biological moments. His advisor tapped him bioinformatics to scientists and
techniques continue to improve. to lecture a course on algorithms clinicians at the UKs National
As a result, the proficiency bar for biology, where Asimenos Health Service (NHS). Part of the
in bioinformatics continues to explained genetics, genomics and countrys 100,000 Genome Project
rise, along with the demand for biology to a class of computer which aims to sequence and
talented bioinformaticians (see scientists. That experience in understand the genome of 100,000
new mobility: a case study). the biological deep end helped patients his centre was the first
A few decades ago the ability to him, because DNAnexus makes of eight to start this educational
scour databases to find a single software for biologists. He needs programme in October last year.
gene provided at least a plank to know the language of the Bioinformatics can often equate
in the platform for a career in companys clients, so he can help to some form of programming.
bioinformatics. Now, that skill is a create software to meet their needs. Although many people are initially
0 1P6R I L 2 0 1 6
basic part of a molecular biologists Bridging the gap between biology intimidated by the prospect of
toolkit, as essential as fundamental and computer science remains one learning programming languages,
wet lab techniques. In response, of his biggest challenges. That Cazier comforts them. His first
bioinformaticians need to keep vocabulary is really deeply rooted lesson is to impart confidence.
improving their skill set. And to in every single discussion, he says. Researchers biologists,
E JAOT BUSR |E 7J OABPSR I| L 7 2 A
really make a mark, they need But even more skills are necessary clinicians, whatever can learn
to develop new tools that others as the technology improves. mathematics and programming,
in the field consider valuable. Knowledge in machine learning Cazier says. To get researchers
"The learning curve is both bigger and artificial intelligence might be comfortable, he uses data from a
and steeper now," says George needed for the next generation of few patients to show how mutations
Asimenos, director of strategic bioinformaticians, Asimenos says. are identified, and then asks them
projects at DNAnexus in Mountain After graduate school at Lund if the mutations are new and if the
4 T0 U | R N
View, California. DNA sequencing University, Sweden, Jean-Baptiste information is statistically reliable.
was relatively slow and expensive Cazier translated his knowledge of I am talking to their research
when he started graduate school 13 applied mathematics into fluency brains, so it works, he says.
C| 0N1 A
RECRUITMENT NATUREJOBS
EVELYN KIING
SPOTLIGHT ON BIOINFORMATICS
Sibon Li
What prompted you to get into bioinformatics? How do you bridge the communication gap between
As a teenager, I was always interested in computers and knew that I computer scientists and biologists?
wanted to do something with them. I would buy PC magazines, and play Attend general conferences rather than going to those that are specialised
with text-based video games. At the same time, at school, there were to your area of research. You get to interact with others outside of
two things that were fascinating to me I didnt care about my biology your field of work and the feedback can be valuable. In addition, other
classes until they taught me about evolution, which was really exciting. researchers may see the significance of your research towards their own,
And the other was probability theory in statistics, where I enjoyed the and might want to collaborate.
problem solving.
Also, try to interact with colleagues across departments. When I first
At the end of high school, I had no idea what I wanted to do at university. started my graduate work, my desk was in the biology department. I made a
I was flicking through a university prospectus and stumbled across lot of friends there and attended biology seminars. Often, I would see how
the bioinformatics program they were offering for the first time at the my research would benefit others and assisted some of my peers in their
University of Auckland. I had no idea what it was or the job opportunities computational work.
in the field but it seemed like the perfect union between all of the things
that I was interested in. What advice do you have for biologists or computer
scientists considering bioinformatics?
Whats your research background?
I think an understanding of bioinformatics and proficiency in computational
In academia my research focus was in developing algorithms for analysis is essential towards being a biologist in this day and age. The
understanding the variation in rate of molecular evolution. As part of impression that I get from biologists is that learning these sorts of
my research, I worked on the BEAST project, which my PhD and postdoc techniques is difficult and outside of their comfort zone. In reality, its fairly
advisors Alexei Drummond and Marc Suchard had developed at Oxford. easy to understand and just requires a change in mindset.
Currently, I work at Google as a software engineer on the Knowledge For computer scientists, I would say that there are a ton of interesting
Graph team, focusing on natural language understanding. Our problems inside biology that are worth solving. The problems in biology are
O RB ES J |O 7B SA P| R 7I LA 2P0R1 I6L 2 0 1 6
technology is used in Google Search, as well as a range of other no different to the traditional problems that computer scientists generally
products. focus their efforts on, in the sense that they are complex, intangible and
challenging. If anything, the benefit to the world is potentially much greater
How did you interact with more biologically-minded
than many of these other fields.
people to solve problems?
Frankly, there was very little interaction between the groups I worked Finally, what does a degree in bioinformatics get you?
with and traditional biologists. Computational biologists know enough A degree in bioinformatics provides you with a diverse skill set that opens
about the biology to find some problems to solve, but often fail to doors for a range of career options. I myself transitioned to working at
address the biologically relevant and interesting problems. Google on pure computer science problems such as natural language
On the other hand, I feel biologists believe that they have all the tools understanding and developing infrastructure. Peers from my program at
| UNRAETJ U
necessary to get good results. Of course, this is inefficient in many university have found work in areas like biostatistics, pure statistics and
cases. This is a general problem in the bioinformatics community there biology. Outside of academia, there are plenty of careers for bioinformatics
needs to be more communication and collaboration across the spectrum. graduates in companies doing things like software and biotech.
1 0| 0N1A T
C
RECRUITMENT NATUREJOBS
That breaks the ice for But if someone with a biology multiple programming languages,
ELISABETH FALL
basic programming especially background doesnt know how he had time for little else. Mullen
when he demonstrates how code to use it, that tool is useless. The estimates he averaged about five
can help them ask and answer two sides need to work together to hours of sleep during his MSc year.
scientific questions. He has been develop user interfaces, she says. The sacrifice paid off, though.
SPOTLIGHT ON BIOINFORMATICS
surprised at the response. I was There are formal efforts The Engineering and Physical
quite worried about the teaching in place to achieve this. For Sciences Research Council
course, but they embraced it and instance, the Global Organisation (EPSRC) and GlaxoSmithKline
asked for more, says Cazier. for Bioinformatics Learning, are funding his PhD research. In
Basic bioinformatics skills can Education & Training (GOBLET) return, Mullen contributes to drug
empower biologists to make use of helps each side learn the others discovery work for the company.
their own data: after all, they have science. But even that can only He has already been offered a
the best understanding of biological go so far. It is unrealistic for government-funded joint postdoc
processes. However, because the specialists from each side of the position with Newcastle and
field is advancing so quickly, they field to be completely fluent Prozomix a biotech company
need to keep in touch with the in the others field. Each side based in Northumberland, United
hardcore bioinformaticians to has to understand their own Kingdom even though he hasnt
have any hope of keeping abreast limitations, Schneider says. yet written up his dissertation. Atul Butte
with the latest developments. David Martin, a senior lecturer Federico Abascal was similarly
in bioinformatics at the University computationally illiterate when
of Dundee, agrees with Cazier that he completed his undergraduate
MATTHEW WAKEFIELD
biologists need more familiarity degree in 1998. Now he works thought combining computers
with bioinformatics. The core as a bioinformatician at the and medicine would prepare
skills for modern data-rich biology Wellcome Trust Sanger Institute him for a career in radiology.
are not always there, he says. If in the United Kingdom one He pursued that career
he could, he would teach every of the worlds most renowned by enrolling in an eight-year
biology grad student enough skills bioinformatics hubs. When programme at Brown University,
so that they could do some basic he finished his undergrad studying medicine and computer
programming, read data into a work, he remained interested science. Towards the end of his
file, then be able to manipulate in biology, but knew he didnt studies, gene expression microarray
and process it not enough want to perform experiments. chips were invented, the human
to be a computer scientist, but He took a course in the genome project took off, and the
enough to have the tools to work programming language C, then era of big data in biology was
with the data, he says (see skills went on to graduate school at the born. He emerged with a skill set
spectrum). But money and time Spanish National Biotechnology training him in both worlds.
is always a problem. These skills Centre in Madrid. His drive to He may have been the exception
Vicky Schneider take time to develop and craft, solve problems in evolutionary then, but Butte sees dual training
much like lab skills take time to biology led him to learn more as the new norm. More and more
develop and craft, Martin says. programming languages. Once people come up with both, he says.
However, it can be done, if one you know one language, it is easier In fact, it is harder now to
Vicky Schneider, associate is willing to put in the work, says and easier to learn others. get into top bioinformatics
professor and deputy director of the Joseph Mullen, a PhD student at He advises would-be graduate programs conversant
EMBL Australia Bioinformatics Newcastle University in the United bioinformaticians to get out of in only biology, or only
Resource, at The University Kingdom. After an undergraduate the lab as much as they can to computer science. You have to
of Melbourne, Australia, says degree in biology and with avoid losing perspective. In my demonstrate you know more than
one way each side can learn the little computational experience, case, the best ideas never come a little of both, Butte says.
language of the other is by having he decided that the career in front of a computer, Abascal But knowing enough to use
more conversations. You have opportunities bioinformatics would says. Its important to enjoy your the software may not be enough
to have a minimal common open up would be worth the effort. job and be motivated: the best to excel, Butte says. The point
vocabulary, Schneider says. It was an incredibly steep learning ideas come whilst Im in bed, of being in this field is to develop
More dialogue between users curve, Mullen says. I jumped into or walking my dog, or having new tools, new methods you
and developers results in better it with everything I had. Indeed, a coffee with colleagues. have to innovate. You have to write
tools, she says. For instance, a between coursework, working three The next generation of new code. Focusing too much
O RB ES J |O 7B SA P| R 7I LA 2P0R1 I6L 2 0 1 6
computer science-trained developer part-time jobs to fund his education bioinformatician may well on one technique or one problem
might create a powerful tool. and putting in the hours to learn find the lab and the computer could be career limiting, he says.
indistinguishable. Dual training He advises constant learning
in both fields, as early as undergrad the amount of data keeps growing
education, may well become the and the nature of it keeps changing.
SKILLS SPECTRUM norm, says Atul Butte, director Treat the field with respect, he
There are three essential skill sets bioinformaticians need. Heres where to start. of the University of California, says. If you want
1. COMMAND 2. PROGRAM 3. DATA San Franciscos Institute for tothriveinbiomedical
Understand how Unix Learn Python, a basic Understanding what type of data Computational Health Sciences. informaticsit cant be a casual
commands work. language. Then consider is in different kinds of databases, Butte is a pioneer in training thing. You have to be here
| UNRAETJ U
R, a useful language and how to mine it, is essential. for bioinformatics. In high school, to stay.
for handling statistics. Learning relational database
techniques is another plus.
he was fascinated by National
Geographic covers displaying This content was commissioned and
2 0| 0N2A T
More information about these roles, closing and interview dates and application instructions are available at www.embl.org/jobs.
EMBL-EBI offers world class research and computer facilities in addition to highly competitive pay and excellent benets plus on-site
amenities such as gym, conference centre, nursery and free shuttle buses. We also benet from close ties to the Wellcome Trust
Sanger Institute and the University of Cambridge.
W275111R
RECRUITMENT NATUREJOBS
U274566EL
1123 April
Single Cell Genomics Chromatin Structure
1416 September NEW Malaria Experimental and Function
Genetics 31 October9 November
EMBLWellcome Genome 814 May
Campus Conference:
Proteomics in Cell Biology Bioinformatics Summer
and Disease Mechanisms School
1417 September NEW 1317 June
Genome Informatics
1922 September Wellcome Genome Campus
Hinxton, Cambridge, UK
wellcomegenomecampus.org/coursesandconferences