Auditory perception of reduction in genome diversity as consequence of plant domestication
This project focuses on communicating, in audible format, reduction in genome diversity as consequence of plant domestication. The plants featured are:
Wild rice -------------------- Oryza barthii
Domesticated rice ------------ Oryza sativa
2_ Maize | Corn
Wild corn -------------------- Teosinte (Zea mays ssp. parviglumis)
Domesticated corn ------------ Zea mays
The assumption implicit in this work establishes that when DNA sequence from a domestication gene is sonificated (rendered in audible format), the resulting sound composition derived from the wild species will be significantly different to that from the domesticated one. This is so because repeated cycles of breeding and recurrent selection since the cultivation of cereals around ten thousand years ago, had shaped the genome of plants and reduced their allelic diversity among populations of cultivated varieties of crops relative to their wild ancestors (free of man-dependent selection pressures). The genes selected were:
(a) Shuttering gene sh4 (rice)
(b) Barren stalk 1 gene ba1 (maize)
Because seed shattering in wild relatives of rice results in greatly reduced grain yield, the loss of seed shattering habit is considered one of the most important events in rice domestication. On the other hand, variation at ba1 gene may have had an important role during the morphological evolution of maize from its wild ancestor, teosinte. Because ba1 was shown not to be a direct target of domestication itself, but the entire genomic region containing this gene has reduced nucleotide diversity compared to wild maize, it was interesting to include it in conjunction with sh4.
On the other hand, mutations at ba1 gene had an important role during the morphological evolution of maize from its wild ancestor, teosinte.
Figure 1. Left photograph: seeds of wild rice (Oryza barthii) and domesticated rice (Oryza sativa), respectively. Right photograph: seeds of Teosinte (wild ancestor of maize) and domesticated maize (Zea mays), respectively.
Gene sonification: auditory rendering of genome diversity
Sonification of rice sh4 gene comprised DNA sequences from 17 different individuals for Oryza bhartii (wild rice) and 20 different individuals for Oryza sativa (domesticated rice) as suggested by Dr. Jae Young Choi from Dr. Michael Purugganan's lab at NYU. The data provided was downloaded from two studies previously published and referenced in the following two publications:
By applying a sonification approach similar to the one previously described by Calvino (2017) in
I was able to generate a sound composition for the entire alignment length of 1,185 base pairs for each population (wild and domesticated sh4 sequences, respectively). I proceded then to render the audio on a multimedia computer and recorded about 10 minutes of sound using the Zoom H4n pro audio recorder, placing the resulting audio files on Adobe Audition and selected and audio segment of less than 2 minutes that displayed remarkable differences from wild rice compared to domesticated rice.
Audio 1. Allelic variance at sh4 gene sequence was sonified and an audio file corresponding to the segment displaying drastic differences was recorded. The first 9 sequences for each species were used to address differences in sound diversity between wild and domesticated rice. When playing the above audio files, striking differences can be listened starting at 30 seconds into the sound composition.
In the audio segment highlighted, there is a moment in which the sound is more discrete, punctuated, and less diverse in domesticated rice as opposed to wild rice. The difference in sound pattern at sh4 is also evident when visualized in graphic form.
Figure 2. Visualization of sh4 sonification by drawing amplitude levels for each sound composition relative to the audio files presented in Audio 1. Amplitude levels at any given point across the audio composition were drawn as colored dots (light orange for wild rice, and pink for domesticated rice) starting from the left (beginning of audio) and ending on the right (end of audio). Amplitude level is graphed along the y-axis with higher values corresponding to dots drawn further down the graph.
I then proceded to apply a similar approach to the sonification of ba1 in maize landraces (early domesticates) and wild maize (teosinte Zea ea mays ssp. parviglumis) for the promoter region located between 1.5 to 2.0 kbp upstream of the transcription start site (region 'c' from Figure 4 on Gallavotti et al. (2004) Nature (432): p630).
Based on Gallavotti's supplementary table 1, I could retrieve the nucleotide sequences for landraces and teosinte individuals. I performed a multiple sequence aligment containing 18 sequences (9 lanraces : 9 teosinte sequences, respectively) in order to sonify a comparable sequence length and thus include any deletion/insertion among sequences. From the entire alignment that included about 610 bp total, I selected a polymorphic region of about 60 bp to sonificate (Audio 2).
Unlike rice sh4 sonification, I could not find a drastic audible differences among nucleotide diversity of landraces and wild maize accessions. This finding can also be viewed in graphical form (Figure 3). Interestingly, I obtained similar results when sonification of ba1 'region a' at the 3' end of the gene was conducted (data not shown). This can be explained based on distributed reduction in nucleotide diversity across the entire region instead of punctuated nucleotide changes and allelic variation among landraces and teosinte sequences.
Audio 2. Allelic variance of ba1 promoter region (region c) among maize landraces and wild maize. A polymorphic fragment of 60 bp was sonified but striking differences in sound among group of sequences could not be found.
Figure 3. Visualization of ba1 sonification by drawing amplitude levels for each sound composition relative to the audio files presented in Audio 2. Amplitude levels at any given point across the audio composition were drawn as colored dots (light orange for teosinte, and pink for maize landraces) starting from the left (beginning of audio) and ending on the right (end of audio). Amplitude level is graphed along the y-axis with higher values corresponding to dots drawn further down the graph.
Artistic interpretation of reduction in genome diversity: ARTE GAGAISTA
I wanted to approach the creation of artworks based on concepts of reduction in genome diversity and plant domestication to convey scientific knowledge in artistic form. This is an additional step within a long term effort in combining genomics with creative coding and abstract art in a new form of artistic expression that I've called GAGAISMO (or ARTE GAGAISTA in Spanish). GAGAISMO stands from Geometric And Genomic AbstractionISM and I am highly interested in its conceptual development (its discourse and theoretical premises) as well as its artistic implementation and practical implications in the world of art, science and emerging media.
Figure 4. Artworks based on rice sh4 gene (top piece) and maize ba1 (bottom piece), respectively
In both of these artworks I wanted to play with biological concepts of DNA and protein sequence text as well as images of 3D protein structures integrated with geometric visual elements. Both pieces contain images of wild and domesticated seeds and in doing so organized the story from the gene-to-protein-to-seed. If domestication involves the act of recurrent selection based on plant characteristics that facilitate agriculture and human consumption of food, then art-based interpretation of plant domestication could facilitate human consumption of scientific knowledge. This means that we could domesticate science concepts for art.
Experimental audio-visual piece based on sh4 sonification and artwork
I created an experimental audio-visual performance as first attempt to incorporate performative and interactive aspects into the communication of science-based concepts in artistic form. For this, I used and implemented concepts from computer vision, server-client network, and creative coding to manipulate audio files previously shown in Audio 1 and 2 (see above)by moving a plant leaf in front of my laptop's camera. The movement of the leaf dynamically altered in real time visual aspects of the artwork shown in Figure 4 (rice_sh4 art piece). I achieved this by creating two independent programs that 'talked to each other' within the same computer, with one program managing computer vision and reactive sound manipulation that sent data to a second program that managed visual manipulation of artwork (Video 1).
Video 1. A computer vision algorithm recognizes (and tracks) the green color of the plant leaf, and as I move the leaf around in front of computer's camera it manipulates sh4 sonification in real time (audio files from wild and domesticated rice, respectively) by altering the volume and rate at which the audio is rendered. The positioning of the leaf also influences in real time the visual algorithm that manipulates aspects of the artwork. Ideally, when performing I should use a rice leaf instead.
The performance intends to bring forward the idea of re-purposing science-based data and present it in artistic form, and thus 'domesticating it' for art purposes.
A physical interface to access auditory data ('the genome audio player') is pending final construction.
I want to thanks Dr. Michael Purugganan and his team members at NYU for providing rice DNA sequences and seeds; to Dr. Andrea Gallavotti from Rutgers University for providing teosinte/maize seeds and DNA sequences of ba1 gene.