The Half Decent Pharmaceutical Chemistry Blog

As many adolescences, at the beginning of high school I went through my rebellious period: interested in anarcho socialism and communism, curious about the “No-Global” movement, committed to many social issues such as immigration, pollution, equity, etc. If you are European, there’s a chance you, too, spent a bit of your adolescence dressing up like a squatter, participating in boring meetings and attending endless and pointless conferences.
That was a long time ago, when hormones can make a male teenager go crazy and oddly embark on silly projects for emulation or a physiological need to do something radical and, in his view, shocking.

Although adolescence, like acne, only lasts a few years and, fortunately, (most of) lads grow up from many points of view, this means I’ve turned myself into a globalisation-enthusiast, who likes fast-foods and famous sportswear brands, with no sense of social commitment (whatever that really means). This said, I just want to make clear I hoped Boris Johnson to become the new mayor of London.

However, my political views will be fully discussed in the Euchromatic Blog, as this post is solely about bioinformatics and, in particular, sequence alignment. To be honest, I can’t stand bioinformatics: I hate having to waste time in front of a computer doing BLAST searches on Gene Bank  or fishing for isoforms on Swiss Prot. Even worse, I’m not particularly excited by people who choose to carry out their research sitting at a computer instead of handling pipettes and getting dirty and smelly with media and reagents.

Unfortunately, last year, my current group leader insisted that I had to take a bioinformatics exam. The idea was to make practice with data bases but, sadly, the course annoyed me with depressing lectures on how Google works or the characteristics of an algorithm.
Nevertheless, the bit about sequence alignment was quite interesting. There are two levels of alignment: an algorithm can perform a pair-wise or multiple alignment. For example, BLAST utilises multiple alignments as it compares our target sequence to all the strings present in a data base.
More interesting is the distinction between global and local alignment: with the former you basically align the whole sequence(s), while the latter is based on dividing a string into segments and it is obviously the ideal option for a data base.

A global, pair wise alignment requires gaps in between the 2 sequences in order to obtain the best score. A global, multiple alignment is what you need when trying to predict secondary and tertiary structures.
A multiple, local alignment, on the other hand, aims at the so-called Maximal Segment Pair (that presents the highest matching score).
Therefore, the algorithm performs a multiple alignment, locally matching and retrieving only those with a score above a threshold.

Great stuff when you cannot sleep, eh?