From Protein Data Bank to... Blender!

Visual Molecular Dynamics (VMD) is a stunning piece of software. Not only it is a powerful tool for molecules visualization and rendering, but it is very useful tool for converting PDB structures into other formats that 3D modeling programs can read. In particular, VMD exports into OBJ, which probably is the most common interchange format.

Moreover, it is usable! A+

Mission accomplished

Rastrigin's benchmark function is far more photogenic than I am

I finally got my Master's Degree in computer science, by defending a strongly interdisciplinary thesis about GPGPU-powered parameter estimation in biochemical systems. The method I propose spins around my favourite bio-inspired optimization method – the Particle Swarm Optimization – whose fitness function relies on simulations produced by Gillespie's stochastic simulation algorithm. But that's just the surface: there's a lot of complex stuff "under the hood" and plenty of room for future developments.

It has been a very interesting experience, enlightening in many ways; the most important aspects came from a human standpoint: I met, and worked with, some wonderful people from whom I've learned a lot, people I'll never cease to admire and be grateful to.

It's kinda funny, anyway: I run away from chemistry at high school, and now I'm onto it again. Strange to see how different routes ultimately lead into the same place.

On the bioinformatics or: computer science's pillars

One of the most interesting facets of bioinformatics is that pushes computer science towards its limit. When you deal with huge data sets, a poorly implemented algorithm or the choice of the wrong data structure leads to inevitable failure.

Screenshot above shows some information about the costs of a piece of software I'm developing for prof. Besozzi, that executes a slight variation of Aho-Corasick on saccharomyces cerevisiae's genome, looking for a bunch of relevant nucleotides sequences. The data-set isn't impressive, just 36MB, and loading it into main memory takes about 10MB. The remaining 128 megabytes are due to the tables produced by algorithm and the huge vectors that store the results (>7 million of matches).

This being the ratio, one can easily foresee that bigger data sets (eg.: the human genome) could easily get out of hand, requiring more resources than one can afford! I'm more than satisfied in terms of time complexity (it's a linear algorithm, therefore optimal), but some further optimization could (has) to be done on the spatial side.