Question

Do you think Matlab is a good tool in biology??

Answer 1

What data processing framework is best depends on what specifically your data is.

For most statistic analysis - stuff like linear regression, model fitting, anova and other hypothesis testing, chi squares, ecology data, epidemiology data, climate data, all that goodness - most biologists use R. R is such a soft language that most people with not much training in computer science (i.e., biologists) use it and believe they are programming; most hard core computer programmers regard it as a statistical suite rather than meriting the status of a real language. But for most sats work, it works superbly and is the common language used by all non programmer scientists in all disciplines.

Matlab is for harder core data processing - especially computationally intensive analysis and engineering applications.

For example, while R has actually have fft() and ffti() for fast fourier transform and inverse fast fourier transform, to implement this in R with any semblance of computational speed you would need to write a Fortran subroutine and get it to run in R, which is a pain to the point of being impractical.

So for computationally intensive work, such as computing FFTs, or for most work involving the interface between biology and more physical sciences - such as collecting and processing NMR data, or processing energy waveforms collected in high tech microscopy - Matlab would be the go to.

There are things I could say about various other languages, and I will say in brief because they are the lesser used/more specialised options:

1. Systems Biology Markup Language is sort of a bio specific implementation of HTML - those already familiar with HTML obviously like it, and it is better than its competitors for some certain tasks, often involving signalling networks and other kinds of networks.

2. Perl and Ruby are both considered good for specific things - though in all candour, every time I talk to a postdoc working in either of those two and ask them why, they tell me that they use it because their PI told them to. And most professional programmers at biology labs and institutes look at these languages and shake their heads.

3. Lisp/Scheme - I have seen and participated in some work done in machine learning in biology - such as training a computer program to predict how proteins fold by having it 'watch' human experts - and Lisp is the language for all artificial intelligence and machine learning work. Again, highly project specific but the only good option in that corner of the field.

4. Python and its packages - scipy, numpy, etc - are sometimes used, I think largely because biology undergrads take one term of it to satisfy their computer skills requirements, and it is the language they all speak. And possibly the same argument could be made about C++.


So depending on what your work and skills are, I suggest you look at R. Particularly for statistic work of any sort. If your work is closer to the engineering or physical sciences side of biology, or if it is computationally intensive, I would go with MatLab. And if you are spectacularly familiar with and skilled in any of these other languages, go with it instead.

Answer 2

I think Matlab is more useful for engineers and mathematicians as a modeling tool for functions and things of that nature.