In this study, we address the challenge of consistently following emotional support strategies in long conversations by large language models (LLMs). We introduce the Strategy-Relevant Attention (SRA) metric, a model-agnostic measure designed to evaluate the effectiveness of LLMs in adhering to strategic prompts in emotional support contexts. By analyzing conversations within the Emotional Support Conversations dataset (ESConv) using LLaMA models, we demonstrate that SRA is significantly correlated with a model's ability to sustain the outlined strategy throughout the interactions. Our findings reveal that the application of SRAinformed prompts leads to enhanced strategic adherence, resulting in conversations that more reliably exhibit the desired emotional support strategies over longer conversations. Furthermore, we contribute a comprehensive, multibranch synthetic conversation dataset for ES-Conv, featuring a variety of strategy continuations informed by our optimized prompting method. The code and data are publicly available on our Github. 1

Bypassing nature's evolutionary trajectory, de novo protein generation - defined as creating artificial protein sequences from scratch - could enable breakthrough solutions for biomedical and environmental challenges. Viewing amino acid sequences as a language, we demonstrate that a deep learning-based language model can generate functional artificial protein sequences across families, akin to generating grammatically and semantically correct natural language sentences on diverse topics. Our protein language model is trained by simply learning to predict the next amino acid for over 280 million protein sequences from thousands of protein families, without biophysical or coevolutionary modeling. We experimentally evaluate model-generated artificial proteins on five distinct antibacterial lysozyme families. Artificial proteins show similar activities and catalytic efficiencies as representative natural lysozymes, including hen egg white lysozyme, while reaching as low as 44% identity to any known naturally-evolved protein. The X-ray crystal structure of an enzymatically active artificial protein recapitulates the conserved fold and positioning of active site residues found in natural proteins. We demonstrate our language model's ability to be adapted to different protein families by accurately predicting the functionality of artificial chorismate mutase and malate dehydrogenase proteins. These results indicate that neural language models successfully perform de novo protein generation across protein families and may prove to be a tool to shortcut evolution.

Evolution-based deep generative models represent an exciting direction in understanding and designing proteins. An open question is whether such models can represent the constraints underlying specialized functions that are necessary for organismal fitness in specific biological contexts. Here, we examine the ability of three different models to produce synthetic versions of SH3 domains that can support function in a yeast stress signaling pathway. Using a select-seq assay, we show that one form of a variational autoencoder (VAE) recapitulates the functional characteristics of natural SH3 domains and classifies fungal SH3 homologs hierarchically by function and phylogeny. Locality in the latent space of the model predicts and extends the function of natural orthologs and exposes amino acid constraints distributed near and far from the SH3 ligand-binding site. The ability of deep generative models to specify orthologous function in vivo opens new avenues for probing and engineering protein function in specific cellular environments.