Arte GAGAISTA from Rice Genomic Segments as Mechanism for Creative Exploration
GEN – Centro de Artes y Ciencias
Montevideo, December of 2016
In this work I explored the creation of visual elements based on DNA sequence data from rice genomic segments as a valid mechanism for creative exploration.
My objective was the creation of artworks at the intersection of genomic information with art and programming in order to transit new avenues of visual imagery. I’ve recently coined this emergent current of artistic expression as GAGAISM, from Genomic and Geometric Abstraction-Ism (GAGAISMO or Arte GAGAISTA in Spanish) (see previous post).
The questions I wanted to explore were: how visual artworks can be generated from (a) abstracting DNA sequence data contained within rice genomic segments of interest; and (b) modulating the disposition of visual elements in response to non-correlated genome data such as current weather conditions across several cities of the world.
The artistic principle from idea to the final artwork involved the creation of computer algorithms to process rice genome and weather data to produce different visual elements that conformed to the aesthetic sensibility inherent to the artist’s style.
Algorithms were applied for the generation of image, color, shape, and disposition of the different elements.
Rice genome browser from RAPDB as raw material for image generation and data retrieval
When creating this artwork I tried to represent the processes that scientists go through when working with genomic data, and the first is accessing DNA sequence information through a genome web browser. In this particular case I used the genome browser from the Rice Annotation Project Data Base (RAPDB: http://rapdb.dna.affrc.go.jp/).
Figure 1. Home page of RAPDB to browse rice genome and genes
Nucleotide sequence of Black hull 4 (Bh4) gene as raw material for color generation
In order to create the background of the artwork I selected the Bh4 gene as it controls the black color of the seed hull in the ancestral wild rice Oryza rufipogon and Oryza nivara (Feng et al. 2011 - http://www.plantphysiol.org/content/early/2011/01/25/pp.110.168500). The non-functional allele of Bh4 determines the white seed hull present in cultivated rice Oryza sativa.
The color palette was determined by applying an algorithm that took into account the frequency counts of A, G, C and T nucleotides present in the 2015 base pairs of the Bh4 gene locus to set the amount of Red, Green, Blue, and Transparency for a single color.
Colors explored according to nucleotide frequency were:
c1- %A = Red %G = Blue %C = Green %T = Transparency
c2- %T = Red %A = Blue %G = Green %C = Transparency
c3- %C = Red %T = Blue %A = Green %G = Transparency
c4- %G = Red %C = Blue %T = Green %A = Transparency
The application of this algorithm resulted in the following colors:
Figure 2. Color palette generated from nucleotide’s frequency occurrence in the Bh4 gene
I selected color #4 (bottom right on Figure 2) as it matched closely the black color of rice seed hull in Oryza rufipogon. I reasoned then that I should apply and algorithm to tint with color #4 an image taken from RAPDB genome browser displaying the genomic context of Bh4 gene locus.
Figure 3. An image from RAPDB genome browser displaying the Bh4 gene locus was tinted with color generated based on Bh4 nucleotide sequence
I subsequently took the first 1600 base pairs from Bh4 gene and applied an algorithm to display each nucleotide as a line with a specific color from Figure 2 (A=color1, G=color2, C=color3, and T=color4 respectively). The gene segment was then displayed as background element to Figure 3, creating the figure to be used as precursor for setting the background of the painting.
Figure 4. Composite image with Bh4 nucleotide sequence represented as colored lines locate behind tinted images from RAPDB genome browser displaying the genomic context of Bh4 on rice chromosome 4
Application of a feedback algorithm to increase the complexity of the composite image creates the background element of the artwork
I wanted to increase the complexity of the image shown in Figure 4 to the point of distortion so the viewer may experience a familiarity without recognition. There were two reasons that supported this decision:
As we seemingly browse the genome of species of interest during our work, we forget all the effort that took the sequencing of a species’ genome in providing the user with a graphic, interactive, and searchable representation of it. Distorting the composite image was my reaction to the endless process of searching information in the digital age and, at the same time call the attention to the viewer that searching genes has become as usual as searching anything else on the web (like searching a book on Amazon for instance).
As a remainder that even as we work on the genetic improvement of plants, the display of information on a computer screen obeys different laws from those in nature but our desire to manipulate and alter the genome of different species through breeding and biotechnology is as intense as our artistic desire to manipulate and alter the disposition of pixels (rather that genes) from an image.
The generative algorithm applied was based on a positive feedback loop in which a random segment of the image is selected and copied. The copied segment of the image serves as the source of the next segment to be copied. This processed repeated thousand and thousand of times created the complexity shown in Figure 5 and necessary to serve as background image for my artwork.
Figure 5. A new image composition is generated through the application of a feedback loop algorithm to the image shown in Figure 4
DNA sequence of OsSPL16 and osa-MIR156k as raw material for generative typography
I took the nucleotide sequence of the gene OsSPL16 that controls the size and shape of rice grains as raw material for the application of generative text that controls the position and size of characters while drawing. A loss-of-function mutation in this gene is responsible for the slender grains and better quality of appearance in Basmati rice (Wang et al. 2012: http://www.nature.com/ng/journal/v44/n8/full/ng.2327.html).
Generative text containing the entire sequence of OsSPL16 gene locus was drawn on the canvas with white color with the position and size of nucleotides constantly altered according to the position and speed of the mouse while drawing. The experience of drawing form on the canvas using generative text derived from DNA sequence contained in a rice gene whose function is to determine the size and shape of grains gave me a novel perspective on the notion of form.
As the mRNA abundance of OsSPL16 is controlled by miRNA156 in rice, with this interaction conserved in plants, I decided to apply a generative text algorithm to the sequence of MIR156k in order to draw it on the canvas along with OsSPL16. It is interesting to note that the algorithm applied to the DNA sequence of MIR156k is different from the algorithm applied to OsSPL16. In this particular case, every nucleotide from MIR156k was drawn in orange color using a visual rule that determined the orientation of text as I typed it on the computer, with the sequence itself becoming the blueprint for the composition.
I intentionally used different typographic algorithms to visually represent the nucleotide sequences of OsSPL16 and MIR156k genes because it was interesting to highlight the biology of the regulation of OsSPL16 expression by miR156k. While the typographic algorithm applied to OsSPL16 sequence changed the size and position of the nucleotides as I draw them on the canvas using the mouse, the typographic algorithm applied to MIR156k sequence relied on visual rules that altered the orientation of the text while typed on the computer. Furthermore, as microRNA genes are non-coding RNAs, I intentionally typed the sequence of MIR156k every three nucleotides as if it was a protein-coding gene in order to create tension for the viewer with a genetics background.
GC composition of 12 different MIR genes as raw material for the creation of visual elements
Each chromosome of the rice genome (twelve in total) was drawn based on the GC content from twelve different MIR genes sequences taken from the miRBase database (http://www.mirbase.org/cgi-bin/mirna_summary.pl?org=osa) respectively. Pairs of red and orange horizontal lines were generated according to the GC content in each MIR gene sequence and place at the center of the image crossing the canvas from left to right. The MIR gene sequences used were:
I selected nucleotide sequences from MIR genes as they are relatively short in length and highly variable in the nucleotide composition as to produce an interesting visual pattern. It is important to note that the scientific community was not aware of the existence of functional non-coding RNAs until relatively recently, thus the representation of chromosomes based on microRNA gene sequences makes justice to the history of science.
Visual deconstruction of ELF3-1 and ELF3-2 circadian gene sequences
I took the nucleotide sequence from the first two exons of ELF3-1 and ELF3-2 genes. ELF3 is a time-keeper gene that resets the circadian clock and integrate environmental inputs such as temperature and photoperiod to regulate several processes such as flowering time (Zhao et al. 2012: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0043705). The onset of flowering is important for grain production and yield and the selection of ELF3 gene for this artwork certainly complements to Bh4 and OsSPL16 genes that control grain phenotypes.
I visually deconstructed ELF3 genes by creating visual elements based on nucleotide frequencies of A, G and Ts and placing them apart in the canvas. The position of As along the first two exons of ELF3-1 and 2 were displayed using pink lines that originate for chromosomes 2 and 6, respectively. The positions of Ts were displayed with concentric circles of light gray color for ELF3-2 and gray circles of the same diameter for ELF3-1 (ascending from left side of the canvas towards its right side). Although the real location ELF3-2 is on rice chromosome 2, I intentionally placed it on the painting on chromosome 1 as to balance the elements for increased visual impact.
Modulating the disposition of visual elements in response to non-correlated genome data such as current weather conditions across several cities of the world
The positions of visual elements across the painting were subjected to change in response to weather data retrieved in real time from six different cities of the world (New York City, Chicago, Beijing, Tokyo, Quito and Montevideo). The selection of weather data as non-correlated data set to genomic information stored on genome browsers’ databases is of particular interest for the following two reasons:
Although plants are highly responsive to weather conditions, current changes in climate don’t influence genomic information stored in digital format as opposed to the natural world (take into account epigenetic changes in response to vernalization in plants as a concrete example).
Breaking the static nature of the artwork by ‘moving’ visual elements derived from genomic data emulates -from an artistic point of view- the genomic response of plants to environmental conditions
The environmental parameters retrieved in real time every 15 minutes were (1) temperature, (2) humidity, and (3) wind speed respectively. Weather data was accessed through API requests to Open Weather Map (http://openweathermap.org/) and integrated to the artwork in real time.
Some of the generated images resulted as follow:
The creative exploration at the intersection of art using computers as an expressive medium and plant genomics allowed the generation of novel visual forms that have scientific as well as an artistic value.
Keywords: Martin Calvino; computational abstract art; geometric expressionism; GAGAISMO; ARTE GAGAISTA; rice; rice genome browser; ArtScience; reactive art; responsive art; data art