Saturday, May 05, 2007

The Music of Proteins

.:Gene2music Project:.

Microbiologists from the University of California at Los Angeles (UCLA) have converted DNA sequences of human proteins into music. The Gene2Music project has been led by Rie Takahashi and Jeffrey Miller who work at the department of Microbiology, and have found a way to "cram the 20 standard amino acids (the building blocks of proteins) into just 13 notes. The duo focus on codons sets of three adjacent bases that code for particular amino acids.

Hrmmm.....yeah......Anyways this is all a very cool music project and is an interesting new twist on the generative music approach.

You can here some examples of the midi files .:Here:.

via ZDnet


Anonymous said...

OMFG the grammar in the ZDNet article as atrocious!!! It couldn't be much clearer that the guy writing the article wasn't sure exactly what he was writing about... I don't claim to be 'Linguo' or anything, but that was downright hard to read.

This is a bit "yawn" really... At least for musicians it is, although for geneticists it could possibly have a greater impact.

OK, so they came up with a way to restrict the amount of possible notes which are generated from a sequence from 20 notes, to 13 chords and seven inversions of those chords, which is effectively an attempt to increase of the 'musicality' of the output, or to be more precise, to make the genetic sequence more easily discerned by ear (less random-sounding) due to reduced complexity that comes from a reduced output range.

This was their goal - keep in mind that they say themselves that "The primary goal of this work is to convert genome-encoded protein sequences into musical notes --in order to hear auditory protein patterns--". (my emphasis)

You can certainly hear the patterns in the musical output, if they are present in the sequence. You can hear from the Huntingtin example that it did allow for auditory recognition of the repetition of certain parameters that were specified by the algorithm...But the actual musicality of the output was far from extraordinary, as you can hear from the output of a random sequence when compared to an actual genetic sequence... There's little difference.

This is more of a way of representing the genetic sequence in a more easily perceived audio form, than it is about making musical data from the sequences. Think of it as a new form of genetic notation, rather than a conversion of the notation into music.

Anonymous said...

Whoops, that's the wrong button, I wanted to preview :/

Anyway, I also wanted to add that I believe that a far more musical output could be generated by following an arpeggiator-like methodology, instead of using the genetic sequence to generate absolute notes, or in the case of this new method, chords.

The sequences could be used to generate both the arp patterns, and chord sequences which would be used as the input notes for that arp.

This would mean that the output could be limited not only to a set of predetermined chords, but to chord sequences, and the arpeggiation of the chords. This additional restriction would result in a more pleasing to the ear result, but the additional complexity in the conversion algorithm would of course require additional complexity in the algorithm to convert it back to a textual genetic sequence - meaning, it'd be cool for musicians but les useful for geneticists ;) Although, it could still meet this team's goal of converting to music "in order to hear auditory protein patterns".

Unfortunately I lack the time to implement this concept fully... Not everyone can afford the luxury of university ;) I kinda hope they might read this and try the idea out, as their work is very similar to the basis of this concept. It's cool, but for real musical use, it just needs that extra tweak...