Bioinformatics curriculum

By Grant Jacobs 13/03/2012 12


Those involved in teaching bioinformatics may wish to check out or contribute to the efforts of the ICSB Education Committee who are seeking input to identify a consensus bioinformatics curriculum.

They’ve started a blog for this purpose, including a draft curriculum (.docx file) drawn from an initial survey.

With no disrespect to them the initial survey is limited given the large size of the field, being drawn from just 41 individuals. Nevertheless let’s briefly look at a few points raised.

They show a graph of responses to the question ’What are the most important topics that someone should understand in order to work in the field of computational biology / bioinformatics?’

most-important-aspect-of-bioinfo-to-teach

I have to admit I don’t quite agree that programming and then statistics are the most important.

Let’s presume the student is already enrolled in a biology major. In that context programming and statistics are certainly important practical background that are unlikely to be covered by the rest of their courses.

A key element missing though, that I’ve touched on before, is the theoretical biology that is used in and underpins computational biology. To me that is the most important background. It is not necessarily taught much in ‘straight’ biology courses.

This naturally allies with the established techniques used in the field and with an understanding of how bioinformatics has developed. (Frequent readers here will know I believe it’s important to understand the history of your field.)

They suggest that the curriculum be broken into two broad topic areas:

  • computation, mathematics and statistics (in yellow in the graph above)
  • chemistry and biology (in blue in the graph above)

There was also a suggestion that ISMB’s topic areas be used as the topic areas for a curriculum (see .docx file).

Footnotes

I’d elaborate further, but perhaps readers are grateful I haven’t time! Feel free to add your thoughts in the comments.

Other articles on Code for life:

Teaching bioinformatics at high school

Choosing an algorithm — benchmarking bioinformatics

Research project coding v. end-user application coding

New bioinformatics journal — EMBnet.journal

Literate and test-driven programming (in bioinformatics)

Retrospective–The mythology of bioinformatics

Know the history of your field, be it science or pottery

Reproducible research and computational biology


12 Responses to “Bioinformatics curriculum”

  • Hello.

    I quite agree with you. I don’t think programming and statistics are more important than biology or molecular chemistry (or the contrary).

    But in fact, when you “do” bioinformatics every days, you search in databases, you calculate statistics on data and you program applications to do this quickly. But we sometime forget the objectives which is to test the hypothesis we proposed. And how can we propose hypothesis without theoretical biology ? As a computer scientist, I think I am not well prepared for such questions, that is why I work in strong collaboration with biologists and bio-chemists.

    Bioinformatics is a field where biologists and computer scientists can (must) live in symbiosis. One can give its view to the other and, from this exchange, new ideas, new concepts then new tools can emerge.

    During my phd, I lived in a bioinformatic lab, mostly computer science oriented. We were convinced that our tools and our ideas were of great use for biologists. Now, I live with only (bio)chemists around and I definitely not do the same research. Mostly because I must convince them that my tools are useful… and they show me that it is not always true. I think they opened my sight to the “real” stuff.

  • We were convinced that our tools and our ideas were of great use for biologists. Now, I live with only (bio)chemists around and I definitely not do the same research. Mostly because I must convince them that my tools are useful… and they show me that it is not always true. I think they opened my sight to the “real” stuff.

    You’ve touched on something that is a bit of a hobby-horse of mine.

    In my case I have a ‘true’ double background in both biology (molecular biology / genetics) and computer science. When someone approaches me with a problem, I try encourage them to explain the biology and what the biological problem is that they wish to have addressed, so that the biology drives how the problem is tackled, as I believe it should.

    I sometimes get biologists telling me what they’d like done with their data, but I’ve seen enough people suggest ideas that aren’t really the best for what they want to achieve on closer questioning that I try encourage people to let me worry about mapping the biological question into a computational one (and giving it back in biological terms).

    Similarly there’s plenty of examples of poor bioinformatics in the experimental literature. Ditto from the computing/computer science side of things. I can recall as a student (many years ago now!) reading some excellent computer science that was being offered to biology, as it were, but the biologist part of me was thinking “what pity, this is great a algorithm but unlikely to be much real use”.

    It helps to be able to map the biological question to the theoretical elements that would be used in the computational approach and then to the computational tools (statistical tools, new algorithms, etc).

  • I think the fact that programming is so high probably reflects the desire of bioinformaticians to write software, rather than reality.

    My experience is that sure, biologists need to script things so they can automate tasks, but the need for full-on software engineering is very limited.

    *If* I had a project that required a software engineer, here’s the thing – I’d employ a software engineer, not a bioinformatician.

    IMO there are WAY too many programmers pretending to be bioinformaticians. The emphasis has to be on the biology in computational biology.

  • Readers should free to offer an informal straw poll of what you think are the most important things that ought to be covered in a bioinformatics course 😉

  • What is important is being able to think computationally about biological problems. That absolutely requires the empowerment that learning to program and perform statistical analysis can bring. To a bioinformatician, being unable to program is like being illiterate.

    Yes the biology is important, but what we have to get away from is the silo mentality of ‘this is the biology module, that is the bioinformatics module and over there is the programming course’. We need to integrate the thinking and techniques with the hands on learning of biology. Redesign practicals so they involve pipettes *and* programming.

    This does require a mindset change. Moving from the ‘follow the recipe, write down the numbers and then try and work out what we were doing later’ process that is endemic to students having to think about the practical before hand, actually understand what they are trying to do because they make decisions about what they need to do, and then the write up of whether it worked. Really we need students to write up the practical before they go into the lab, then fill in the blanks afterwards. One of the insights that students gained from a genome sequencing practical we did was that bioinformatics is just as experimental as the wet lab, with just the same issues of uncertainty and need for validation.

    Integration is the first step. Unless bioinformatics teaching is thoroughly embedded in the biology, and students are empowered to make use of the data generated (why do we do the practicals we do? because our students are functionally illiterate when dealing with practicals that actually generate more data points than they have fingers), it will continue to be seen as the domain of the nerds in the basement who couldn’t get a job doing proper biology. Instead of the empowering technique it is that allows them to approach their understanding of life in a new and more powerful way.

  • MIck,

    “I think the fact that programming is so high probably reflects the desire of bioinformaticians to write software, rather than reality.” – that’s a good thought and you might be right. I know I like to be coding, although for me I’m just as happy reading the experimental research literature and putting all the bits together.

    “My experience is that sure, biologists need to script things so they can automate tasks, but the need for full-on software engineering is very limited.”

    This might depend on the target audience for the course – biologists to pick up some bioinformatics as part of a biology course, or a specialist course training those that what to work as bioinformatics researchers. I think the ICSB survey is aimed more at the latter than the former.

    (More later – have to whip up a post to encourage locals to attend an event on tomorrow.)

  • We need to get away from treating bioinformatics as a ‘specialism’ – bioinformatics skills are becoming part and parcel of being an experimental biologist. Everyone needs the skills to a degree and they need to be taught from the start.

    This does not remove the need for bioinformatics specialists.

  • I wonder what the backgrounds of the people sampled was? I find that those who come from a biology background tend to view programming as an opaque discipline. If the people sampled mostly came from that background then an element of that attitude could have been retained i.e. the big thing to move into bioinformatics was learning to code.

  • tirohia,

    Meant reply days ago…!

    You might be right. (The sample is small and might be skewed, too.) Hard to know if it’s that bioinformatics people like coding new things (it’s fun) or if your idea is more where this emphasis is coming from without doing another survey…!

    One quibble I’d add, though, is that adding coding to biology does not add what is, to me, the essence of bioinformatics – the underlying theoretical science, the statistics and so on.

    Adding coding to biology would give a leg up on (basic) data prep. for analysis using existing tools. Bread and butter stuff, and important in that way, but for a teaching course I’d want to see the concepts underpinning bioinformatics analysis/research taught.

Site Meter