Unified encoding of copolymers

Develop a unified encoding strategy for copolymers that represents compositional complexity, variable sequences, and uncertain repeating units while distinguishing copolymers with different sequences.

Background

The paper explains that encoding strategies for homopolymers are comparatively well established, whereas copolymer representations must account for composition, sequence, connectivity, and repeating-unit ambiguity. Existing approaches have important limitations: polymer-fingerprint accumulation cannot efficiently distinguish sequence variations, graph-based encodings require prior knowledge of repeating-unit connectivity, and sequence-based Transformer representations may fail to capture intrinsic sequence dependence. The paper therefore identifies the development of more capable copolymer encoding as an unresolved challenge.