The Wired News piece on Marshall Beddoe’s work with bioinformatic algorithms to crack protocol reverse-engineering has had some good results, in the sense that it’s shedding a bit more light on possible crossovers between bioinf and Net applications. (I’ve written about this sort of genetic cyborg weird-ass mutie crossrbreeding before, for New Scientist).
In the discussion I had with Marshall about his next steps, he commented that a lot of his future work had less to do with number-crunching, and more to do with visualisation. One bioinfodesign innovation that got pointed out to him as a result of the article was sequence logos.
Sequence logos are graphics used in bioinformatics to visually highlight commonalities between multiple sequences. The rows shows how often letters in a gene sequence occur at each position (the GATCs – the taller the letter, the more often the nucleotide appears) together with a measure of how much commonality is preserved over all the sequences (the curve).
If you’re trying to spot patterns in long data sequences, some adaptation of this might be useful to you too. I’m really interested in seeing what Marshall comes up with next.