Biology needs to become prospective

Strategizing the search for bio-utility: A framework for evolution-integrated _in silico_ bioprospecting

MSA-based pLMs encode evolutionary distance but don’t reliably exploit it

Equivalent linear mappings of deep networks are a promising path for biology

A quantitative-genetic decomposition of a neural network

From black box to glass box\: Making UMAP interpretable with exact feature contributions

Efficient GFP variant design with a simple neural network ensemble

Mapping the spectrum of archaeal protein sequence–structure relationships

G–P Atlas\: A neural network framework for mapping genotypes to many phenotypes

Do protein language models understand evolution? Mixed evidence from ancestral sequences and ESM2

Modeling human monogenic diseases using the choanoflagellate *Salpingoeca rosetta*

Modeling human monogenic diseases using the tunicate *Ciona intestinalis*