Developing an automated pipeline for quantifying animal pigmentation using deep learning


Meeting Abstract

47-1  Sunday, Jan. 5 10:30 – 10:45  Developing an automated pipeline for quantifying animal pigmentation using deep learning ALVARADO, SG*; KRUPAKAR, H; Queens College CUNY; GE Healthcare sebastian.alvarado@qc.cuny.edu http://www.alvaradolab.com

Coloration is a salient trait across the animal kingdom that can allow an individual to become cryptic, conspicuous, or social. While some developmental patterns in pigmentation are static, others are dynamic to changes in their ambient environment. Despite a great deal of study in developmental pigmentation patterns, little is known about how environmental cues shape the developmental plasticity that allows an individual to change color. One approach to understanding these processes is through the lens of epigenetic modification and DNA methylation. DNA methylation of cytosine residues in gene promoters is a reversible modification that silences gene function in vertebrates. Since DNA methylation is involved in programming various cellular functions, it is likely that it facilitates molecular changes as pigment-bearing cells (chromatophores) change their composition during animal color changes and behavioral transitions. We used an African cichlid model system (Astatotilapia burtoni) with discrete reversible color morphs (blue and yellow) to dissect the underlying molecular processes that lend plasticity to animal coloration. Our findings suggest that epigenetic processes such as DNA methylation lend plasticity to coloration, which is an important hallmark driving selection. Furthermore, since genetic diversity does not account for the phenotypic diversity seen in Lake Tanganyika, we propose that DNA methylation may contribute to the processes that have led to the adaptive radiation of cichlids in East African Great Lakes.

the Society for
Integrative &
Comparative
Biology