Bioinformatics, biostatistics and computational biology

Question

I am wondering how others define the boundaries between bioinformatics, biostatistics, and computational biology.

In academic setting these three are treated as more or less equivalent terms or as significantly overlapping ones: genome analysis, phylogeny reconstruction, evolutionary models, modeling epidemics, calculating RNA structure, etc. All of these activities fall under any of the three titles with perhaps a slight accent on heavy data load when talking about bioinformatics or more of statistical analysis when talking about biostatistics.

However, when looking for a job in industry, the things change:

• Bioinformaticians are largely expected to be able to deal with large quantities of raw data: cleaning, assembling and doing a bit of analysis;
• Biostatisticians are expected to have a perfect knowledge of the experiment design and the multitude of the statistical tests;
• Computational biologists are those who do not have a proper (i.e. confirmed by a diploma) bioinformatician or biostatistician training, but came to the field from elsewhere (biology, physics, etc.) Although this largely prevents them from competing for bioinformatician and biostatistician jobs, big research labs are interested in such people.

This reflects my own experience and observations, which is why I would like to know opinions of the other members of the community.

Answer 1

Those look good definitions and I would intuitively feel the same. One of the major hurdles I have is between epidemiological/mathematical modellers who work with disease instance data and those who work on molecular data. The latter we call bioinformatians the former mathematicians. The problem is that we could actually work together, but in practice we never do. I understand their stuff, they understand mine ... but it never really happens. It is a limitation.

Personally I would carry the definition through the ranks, so a statistician vs a biostatistician the later understands in vitro experimental data and ideally molecular data (which is harder for them). However, the actual techniques are very similiar and statisticians can adapt to molecular problems.

Personally I would see computer biologists as 'hardcore' bioinformaticians because of the speed at which they can solve problems in code.

The major problem in biology is its tribulism. Python vs Perl is one example. For intermediate level coders (like me) it makes a difference to the speed (and possibly) the robustness of a solution (like pandas is a great 'walking stick' for database solutions). However, the reality is its really about how good you are with code and there are some very strong coders on this site. Ultimately code is code.

Answer 2

A bit off-topic and more of a long comment, but still relevant is computational biochemist.

I and others call myself a computational biochemist and go to the bother of correcting others when called bioinformatician as I do not see myself as one in the strictest sense, but the fact I find myself answering Qs here means that the definitions are loose.

It is tribalism as @michael mentioned, but it is also separate islands. I have as much of an idea what an autozygous BAM file on an IGV plot is, as a genetics bioinformatician has about Lennard-Jones potentials. It is more cheminformatics applied to biochemistry than genetics bioinformatics applied to biochemistry. Unsurprisingly, given that biochemistry and chemistry overlap heavily —particularly in the structural and analytical subfields.

Definition

Computational biochemistry is the use of computational methods to investigate structural properties of proteins or nucleic acids. Namely studying 3D structures, generally utilising forcefield models, for example:

• enzyme/theozyme engineering (not much of that in BioInfo SE as its mostly on Rosetta Forum —but I'd like to change that...)
• docking
• de novo structure prediction
• other structural predictions (mutations, PTMs etc.)
• MD simulations

For a real example, myself (gulp), I normally work on structural effect predictions —geneticists finds a mutation and wants to know more, and depending on what the effect is I will run some analyses, for example for G-protein β-subunit mutations at the interface I modelled the mutations and analysed how it affected the binding energy with the different partners. Possibly ironically given this post, I made a visualisation tool aimed at making it easier for geneticists to share and look at protein structures. During Covid I am working on docking as part of the Covid moonshot, but previously I was worked in enzyme engineering.

New tags

However, I am in half a mind to create a tag computational biochemistry and cheminformatics once I get the wiki privilege. If I do, should I retroactively edit it on past Q&As?