Opiniomics

bioinformatics, genomes, biology etc. "I don't mean to sound angry and cynical, but I am, so that's how it comes across"

On bioinformatics, and torturing the mechanic metaphor

I acted as editor and reviewer for a recent opinion piece in Frontiers titled “Who qualifies to be a bioinformatician?”.   Whilst many find this navel-gazing boring, I do understand the need for bioinformaticians to define themselves, with so much expected of us, from setting up BLAST servers, administering websites, carrying detailed statistical analysis, supporting core facilities, supporting researchers, building databases, training others and fixing the bloody printer.

In the above paper, the following section seems to have hit home, with a few people picking up on the mechanic metaphor:

Bioinformaticians are scientists who develop and conduct research based on a bioinformatics approach, they do not just use the tools to better understand a biological problem. It is a little like saying that driving your car to work does not make you a mechanic.

The message is clear.  If “all” you do is use bioinformatics software, then you are not a bioinformatician; equally, if “all” you do is drive a car, it doesn’t make you a mechanic.  I can see what the authors are going for here, and I agree, but only to an extent.

You see, a mechanic can take an existing car, and they can fix it; replace parts. They can tweak it and improve performance.  But they can’t build you a car from scratch.  So we can extend/torture the mechanic metaphor and say “just because you can fix a car, it doesn’t mean you can build one”.  So perhaps “mechanics” are like pipeline builders in bioinformatics.  They can put things together, but ask them to create something new, and they are (often) lost.

We can go further.  The people who build cars are not those who design them.  That’s a completely different set of skills.  Of course in software, design and build are often (but not always) carried out by the same people, but as we all have seen, that can have disastrous consequences, resulting in sophisticated software that no-one can use.  We may wish to think harder about separating design from build in bioinformatics, there are benefits.

Beyond the designers are the engineers, who figure out how the car will actually work.  Improving on the original design of the internal combustion engine; improving performance; engineering new parts and new techniques that will make the car better, quicker, safer, more powerful or more economic.

So which type of bioinformatician are you?  Engineer, designer, builder, fixer (mechanic) or user?  Oh wait, I forgot, the “user” isn’t a bioinformatician.  So what are they?  Hello “Data Scientist”!!

17 Comments

  1. There is a prize for the 1st person to comment: “ah but you forgot about “

  2. I have a vague memory of writing an article about this for Nature Biotechnology with you.

  3. To stay with the cars and mechanics analogy, here’s my take:

    The bulk of ‘bench biologists’ are passengers that need to be taken somewhere sensible with their data. Most often, the rely on a ‘computational biologist’ to drive them there. The computational biologists have a pretty good sense where the biologists want to go (biological mindset) and a sense of secure driving (appropriate statistical tests etc.). The computational biologists also have at least an idea why their taxi may sometimes stall. However, they primarily use the car. They aren’t particularly passionate about the technology, but they can talk to somebody to change the oil (install BLAST on a server) or call the engineer to point out missing features (doesn’t work with X-seq). That brings us to the people who engineer and build the cars. Those are the ‘bioinformaticians’ , but they as well come with different qualities, reaching from hands-on pipeline builders (software engineering mindset) to theoreticians who deal with the aerodynamics of that car (de Brujin-graph fetishists).

    So did I forget anyone?

  4. We once somewhat flippantly used the analogy heating and plumbing for the data coordination work we do but it does fit quite well with the rest of this

  5. I always thought “bioinformatician” was a dumb term. Why is biology always trying to find new classifications for everything? A physicist can either perform experiments, write code, do complicated maths… There is no such thing as “physico-informatician”, “physico-mathematician”, “computational physicist”, … Why do we need those terms in the first place?

    • Agreed, but note there is also the field of cheminformatics

    • ryan_banana_guy

      5th May 2015 at 2:53 am

      Always thought if you study life you are a biologist. That covers everything, yet people endlessly want to rename departments, titles, and disciplines–integrative quantitative systems computational bio bigdata science. Nobody on the inside or outside really knows what any of those are, so it’s really a disservice to constantly be trying these. Just stick with bio and then explain your skills/approach with additional words. Really no need for ‘bioinformatician’ in my opinion, you’re either a biologist that uses or develops bioinformatic tools, or else you’re a CS or software engineer who works on developing/optimizing tools for biologists.

  6. Following up on what I think is the truly interesting point of discussion here, your suggestion that “We may wish to think harder about separating design from build in bioinformatics, there are benefits.” That’s something we have some experience with in the context of GATK development. For several years we have already benefited greatly from bringing together computational scientists (including but not limited to computational biologists) and “pure” software engineers. The comp sci team members develop the methods/algorithms and produce the initial implementations (~design), and the soft engs help build out the underlying infrastructure and “productionize” the method implementations into hardened tools (~build). This core dev team interacts with analysts (some scientist end users, some people from the genomics platform’s tech dev labs) who provide ongoing feedback about various aspects of the tools and workflows and help define what enhancements and new capabilities are needed, in step with the evolution of technology and research priorities. This is a model based on what actual software development companies do, which we’re further developing through some recent reorganization and the formation of a new entity within Broad called “Data Science and Data Engineering” (see https://www.broadinstitute.org/dsde/).

    Perhaps ironically, the only person in all of this to be called a bionformatician (actually, bioinformatics scientist) is me, and I don’t really do any of what you originally described. Which brings me to the part where I say “ah but you forgot about the driving instructor who teaches non-specialists how to drive the car” 😉

  7. I’ve often thought of this distinction in similar ways there are bioinformatics savy biologist, bioinformaticians, and computational biologists. This would map roughly to car aficionados, mechanics, and mechanical engineers (using your analogy).

  8. In the current day and age, driving your car to work does not make you a mechanic, but until recently, you did need to have some of the skills of a mechanic to keep your car on the road. 50 years ago a car owner would pretty much need to be able to strip down their engine. Even twenty five years ago, car owners would have needed to be able to replace the spark plugs and change the oil, many would have been able to change the cam belt and adjust the timing. You might not have wanted them working on a race car, but they could keep their own commuter car on the road.

  9. Nice post, I’m actually thinking about studying bioinformatics in Ph.D.

  10. I am the first author of the opinion piece discussed here. I follow, since the beginning, all the buzz on twitter and blogs about my letter. I am incredibly surprised by all the reactions and by persons qualifying my letter as “Navel-gazing”. My goal was and is still not to include or exclude anybody from the bioinformatics field.

    The problem is that many scientists in the biological field simply perform a couple of blast on the NCBI website and qualify themselves as bioinformaticians. And so what you will say? If every person performing a bunch of blast could add in their CV that they are bioinformaticians in addition to a lot of “wet-lab” skills, it will be hard for “real” bioinformaticians to stand out. A bioinformatician is able to use tools AND to understand really how they work. A bioinformatician not just uses blast, he or her understand how it works.

    Moreover, as indicated in the opinion letter, it is important for job descriptions (for the bioinformaticians and their employers). Bioinformaticians are not informatics technicians!

    It is clearly indicated in the letter that gray zones exist. I do not pretend to be able in 1000 words to cover all exceptions and to go deep in the subject. Many persons can be considered as bioinformaticians without fitting perfectly in the definition provided by the letter.

    Finally, the main goal of the letter was to start a discussion on this neglected question of “Who qualifies to be a bioinformatician?” It seems obvious that it is a sensitive topic based on the reactions. So the next question would be why it is so delicate?

Leave a Reply

Your email address will not be published.

*

*

code

© 2017 Opiniomics

Theme by Anders NorenUp ↑