Computational Blue-Melanin Design
A TUCCA project applying deep-learning protein design to sustainable colorants — designing melanin pigments in colors that do not exist in nature.
The challenge
Section titled “The challenge”Natural melanin, polymerized from tyrosine (Y), is a robust pigment but exists only in a narrow range of colors (yellow to brown) because controlling melanin size (length) is difficult. Extending melanin to new, stable colors — blue, purple, green — with sequence-to-color tunability is an open design problem.
The computational approach
Section titled “The computational approach”This project uses RFdiffusion to motif-scaffold a pentapeptide-repeat protein (PRP), narrowing roughly 160,000 possible combinations down to a library of 905 Y-containing pentapeptides selected for two properties: tight packing during peptide assembly and high solubility in aqueous environments.
Two of the most soluble designs formed stable blue melanin with λmax absorbing at 615–620 nm — driven by homogeneous melanin length around 60 Y units. Other designs produced purple and green (and the more familiar red, yellow, and brown). The blue melanin showed thermal stability at autoclave temperature (121 °C) and photostability over weeks under illumination, and was demonstrated as an electrophoretic ink.
De novo color design from simple peptides could change how colorants are sourced and produced, and points toward deep-learning tools that predict color directly from amino acid sequence.
Read the preprint
Section titled “Read the preprint”Linked external resources are independent of TUCCA and Tufts University and remain under their own licenses.