ESM-2
PulseAugur coverage of ESM-2 — every cluster mentioning ESM-2 across labs, papers, and developer communities, ranked by signal.
6 day(s) with sentiment data
-
New PIT-GCL framework uses topology for protein interaction prediction
Researchers have developed PIT-GCL, a novel framework for predicting protein interactions that leverages topological graph contrastive learning. This dual-tower system encodes proteins independently using sequence embed…
-
Drug-target affinity model performance varies with evaluation shifts
A new research paper explores how different evaluation methods for drug-target affinity (DTA) models can lead to varying conclusions about model performance. The study found that when models are tested on data distribut…
-
New framework probes protein foundation models for biological interaction data
Researchers have developed ORBIT-FMIB, a new framework designed to analyze how protein foundation models like ESM-2 process different types of biological interaction information. The framework uses Walsh-based decomposi…
-
New latent protein languages boost autoregressive generation models
Researchers have developed two novel latent protein languages, Protein Latent Language (PLL) and Structure Latent Language (SLL), designed to improve autoregressive transformer models for protein sequence and structure …
-
New methods boost efficiency in protein language models · 2 papers
Two new research papers explore methods for making protein language models (PLMs) more efficient. The first paper analyzes the impact of quantization and parameter-efficient fine-tuning techniques like QLoRA on various …
-
Lightweight Murmur2Vec embeddings match heavy PLMs in biological classification
Researchers have developed Murmur2Vec, a lightweight and efficient embedding method for biological sequence classification that rivals the performance of larger, computationally intensive protein language models (PLMs) …
-
New deep learning model predicts antibody-antigen binding affinity from sequences
Researchers have developed DuaDeep-SeqAffinity, a novel deep learning framework designed to predict antibody-antigen binding affinity directly from amino acid sequences. This method bypasses the need for costly and scar…
-
Peptide-protein affinity models benchmarked across diverse data shifts
Researchers have benchmarked various peptide representations and regressors for predicting peptide-protein affinity, revealing that model performance varies significantly depending on whether the evaluation involves shi…
-
New method uses geometry to interpret protein language model features
Researchers have developed a new method to interpret the latent features within protein language models (pLMs) by using geometric annotations of protein backbones. This approach, applied to the ESM-2 model, reveals that…
-
Discrete Ricci curvature offers lightweight protein fold classification
Researchers have developed a novel method for protein fold classification using discrete Ricci curvature on protein contact graphs. This approach generates a lightweight, 22-dimensional feature vector that outperforms l…
-
Canopy model advances metabolic engineering with heterogeneous graph foundation
Researchers have introduced Canopy, a novel heterogeneous graph foundation model designed for metabolic engineering. This model integrates diverse data sources, including genes, proteins, metabolites, and experimental r…
-
Protein language models fail to recover allergen epitopes, study finds
A new study has found that residue-level attributions in protein language models do not accurately recover allergen epitopes, despite the models' robustness in protein-level allergenicity prediction. Researchers develop…
-
Protein language models' allergen explanations lack biological grounding
A new study published on arXiv questions the biological relevance of explanations provided by protein language models used in allergenicity classification. While models like ESM-2 and DeepPlantAllergy demonstrate strong…
-
Sutra language compiles programs into PyTorch neural networks
Researchers have developed Sutra, a functional programming language that compiles into PyTorch neural networks. This system targets vector symbolic architectures by reducing programs to fused tensor-operation graphs. Su…