With an alphabet comprising just four letters, DNA sequence isn’t much to look at. So, when sequence analysis tools want to highlight key elements, they typically do so using colour, font, or by overlaying other types of information. In the not-too-distant future, there may be another option: Audio.
In a paper published this past April in BMC Bioinformatics, molecular biologist and part-time drummer Mark Temple of Western Sydney University, Australia, describes “an auditory display tool” for DNA: sequence in, audio out.
Available online at dnasonification.org, the tool does precisely what it sounds like: Given a sequence of DNA, it will convert the As, Cs, Gs, and Ts into notes played by a virtual piano, guitar, and organ. An ancillary browser extension, called Jazz-Plugin, is required to play the resulting MIDI files, though Temple has made a number of example MP3 files available on his web site and on YouTube.
After uploading a sequence, the user can select precisely how the musical transcription is accomplished. The simplest mode maps each base to a single note, providing a four-tone auditory landscape. Another maps dinucleotides to notes, increasing the complexity to 16 total sounds.
Most informative, says Temple, is the trinucleotide mode. Here, the software maps each nucleotide triplet to one of 20 notes, and outputs the audio in each of three reading frames at once, just as the genetic code maps 64 codons to 20 amino acids. The result is a series of three-note arpeggios – CGF-ADD-CFF-DFG-AFC-GCD-FCD-FCD, for instance. Optional parameters allow the user to flag start and stop codons, or to cause audio in each reading frame to turn on and off as start and stop codons arise.
Genome browsers, Temple explains, typically use visual cues to make key sequences stand out; audio, he figured, could provide a useful accompaniment. “I’m not saying audio by itself is the bees’ knees for interpreting DNA sequence,” he says, “but surely audio can contribute to your visual interpretation.”
That’s particularly true for repetitive elements, slight alterations in which can be difficult to pick out by eye. In one example, Temple provides a sequence of human telomeres – the highly repetitive sequences found at the end of eukaryotic chromosomes. The “sonified” sequence sounds unremarkable and repetitive until it reaches a single base insertion, at which point the music shifts, as if the musician was inspired to shake things up. (Check out an MP3 of the audio here, and a video below)
Temple compares the information that sonification provides to the way the brain perceives its environment: It’s easier to notice objects that are changing than those that are not. “When you look at a field or something, and something moves, like a bird flies, your eyes are attracted to the bird straightaway; that’s what your brain follows. And with this [software] here, if you can create a landscape that’s got a pattern to it, when that pattern changes it’s really clear to your ear that it’s changed.”
He is not, Temple notes, the first researcher to have this idea. Other researchers have also written algorithms to convert DNA sequence to audio, while modern artists have used algorithms to turn sequence into music stripped of analytical value. But by playing all three reading frames at once, and sonically highlighting start and stop codons, Temple says, his approach adds a novel layer of analytical value.
At this point, Temple says, DNAsonification.org is a mere proof of principle – a demonstration of the power of audio to reveal the hidden complexity of DNA. Among other things, he would like to update the software to highlight the sequence that is ‘playing’ so researchers can relate audio cues to the sequence that created them.
He also would like to work with genome browser developers to add DNA sonification as another ‘visualization’ option in DNA analysis – the equivalent of another ‘track’ in the output. (I asked the UCSC Genome Browser project team whether they would be interested in incorporating such data into their software. Robert Kuhn, the project’s associate director, replied, “We do not expect to include such a data transformation into the Browser. As an analytical tool, the method does not appear to offer any biological insight not already available by other methods.”)
Temple sees value in DNA sonification as an educational tool, as well. Because the site converts DNA to audio in the same 5’ to 3’ direction that the cell uses to read genes, Temple explains, “Understanding how the audio is made helps you understand the biology of how DNA is processed and read in the cell.”
If nothing else, it sounds pretty cool.
Jeffrey Perkel is Nature‘s technology editor