sequence-to-function modeling

Benchmarking Spectral Library Prediction Platforms for Neuropeptidomics Applications

This study benchmarks four spectral prediction platforms for neuropeptidomics, finding no single model captures neuropeptide fragmentation well and highlighting the importance of neutral loss fragments and hybrid libraries.

GFMBench-API: A Standardized Interface for Benchmarking Genomic Foundation Models

Wrap a genomic foundation model's inference methods once to score it on supervised and zero-shot GUE, VariantBenchmarks, BEND, TraitGym and ClinVar tasks; code is promised upon acceptance.

An Artificial Intelligence Model for Translating Natural Language into Functional de Novo Proteins

This paper introduces MP4, a transformer-based model that translates natural language into functional de novo proteins, achieving high performance in sequence realism, fold quality, and function alignment.

Distribution-Constrained Optimization for Reliable ML-Guided 5'UTR Sequence Design

This paper introduces a constrained optimization method for designing 5'UTR sequences using deep learning models, ensuring candidates stay within a trust region defined by an OOD score based on KNN distance.

FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications

FLIP2 adds seven protein fitness datasets and 16 splits for enzymes, light-sensitive proteins, and interactions; simple ridge and zero-shot baselines often rival fine-tuned protein language models.

Coevolution-informed Bayesian optimization for sample-efficient protein design

This paper introduces ALSEBO, a Bayesian optimization method for protein design that efficiently finds optimal sequences with few evaluations by leveraging coevolutionary statistics.

Functional pocketomes and complete allosteric maps of five protein kinases

Energy-model maps of five human kinases show conserved distance-dependent decay of inhibitory allostery, but distal allosteric pockets and activating sites diverge between kinases.

HaloMPNN: retraining ProteinMPNN on halophilic proteomes for salt-tolerant enzyme design

This paper retrained ProteinMPNN on halophilic proteomes to design salt-tolerant enzymes, showing HaloMPNN shifts protein properties toward halophilic traits.

A sequence-to-function model to predict T7 transcription rates and redesign T7 expression systems with lowered production of immunogenic RNA byproducts

Fast remote nucleotide sequence alignment with Riboseek

Riboseek is a fast and sensitive sequence alignment tool for RNA and DNA that uses di-mers for alignment, outperforming BLASTN and nhmmer in speed and sensitivity.

CENO: A Genome-Scale World Model for Evolutionary Sequence Interpretation and Programmable Regulatory Design

CENO is a long-context generative model for genomic sequence understanding and design, showing improved variant prediction and regulatory sequence generation across species.

Accelerating Inference in Genomic Foundation Models via Speculative Decoding

This paper introduces speculative decoding to accelerate genomic and proteomic sequence generation using foundation models without compromising prediction quality.

Steering Sequence Generation in Protein Language Models through Iterative Lookback Monte Carlo Sampling

This paper introduces ILMC, a sampling strategy for protein language models that improves diversity while steering generation toward desired properties.

Theseus: Fast and Optimal Affine-Gap Sequence-to-Graph Alignment

Theseus provides a fast and optimal affine-gap sequence-to-graph alignment algorithm, outperforming existing methods in speed while maintaining optimality.

AINN-Express: A Leakage-Aware, Sequence-Only Predictor of VHH Antibody Expression Built on the AINN-P1 Protein Foundation Model

AINN-Express predicts VHH antibody expression using only sequence data, achieving high ROC-AUC scores in both known and new programs without requiring structural data or task-specific training.

pHaseMD4AI: Phase-Space Dynamics Dataset with Chemical and pH Perturbations for Physically and Kinetically Consistent Biomolecular AI

The paper introduces pHaseMD4AI, a molecular dynamics dataset with chemical and pH perturbations, to support biomolecular AI research and benchmarking.

Beyond the Static Caliper: Dynamical Translocases and the Mathematical Imperative for Single-Molecule Proteomics

This paper proposes dynamical translocases and Physics-Informed Machine Learning to overcome biophysical limitations in single-molecule proteomics, achieving >90% accurate amino acid classification.

BioReason-Pro: Advancing Protein Function Prediction with Multimodal Biological Reasoning

A reasoning LLM combining ESM3 embeddings with GO-GPT term predictions produces protein function annotations that experts rated tie-or-better than UniProt in 79% of cases (SFT model).

Amplicon/Protein Bead Display enables quantitative in vitro biochemistry at scale

The paper introduces APB-Display, a method for high-throughput in vitro protein library production and analysis, quantifying binding affinities and expression levels of thousands of variants.

Machine learning-based prediction of human structural variation and characterization of associated sequence determinants

A sequence-only convolutional neural network and an annotation-based random forest predict human structural variant breakpoints from 400 bp windows (AUROC 0.903, 0.904); their ensemble reaches 0.916.

Context-aware sequence-to-function model of human gene regulation

The paper introduces Corgi, a context-aware sequence-to-function model that predicts gene expression and epigenetic signals across unseen cell types using DNA sequence and trans-regulator expression.

Understanding Language Model Scaling on Protein Fitness Prediction

This study shows that larger protein language models can overestimate sequence likelihood, leading to poor fitness prediction, and recommends moderate model sizes for better performance.

ARSENAL: Learning Transferable Regulatory DNA Representations with Targeted Short-Context Language Models

A 350 bp masked DNA language model trained on ENCODE cCREs with a frequency-domain regularizer recovers TF motifs zero-shot, scores regulatory variants, improves ChromBPNet inputs, and guides sequence design.

ProtAug: An Empirical Investigation of pLM-Guided Data Augmentation for Protein Sequence Prediction Tasks

This paper evaluates pLM-guided data augmentation for protein sequence prediction, showing that it improves performance across tasks with variation levels between 2-30%.

Dendritic Wave Recurrent Neural Networks

This paper introduces the Dendritic Wave Recurrent Neural Network (DW-RNN), which improves upon wave recurrent neural networks by adding dendritic input integration, achieving better performance and stability on sequential tasks.

Computational design of blue melanin with peptide motif scaffolding

This paper designs blue melanin using peptide motifs and computational methods, achieving stable blue color with specific absorption wavelengths and thermal/photostability.

Predicting Antibody Self-Association with Sequence Structure Fusion Models: The Central Role of CSI-BLI in Early Developability Screening

The paper develops a sequence-structure fusion model to predict antibody self-association using CSI-BLI data, achieving high performance in distinguishing high and low self-interacting clones.

Breaking the Synthesis Barrier for AI-Designed DNA Libraries

This paper introduces PGLD, an AI method that designs cost-effective DNA libraries for high-throughput experiments by optimizing synthesis-aware stochastic libraries.

Beyond native sequence recovery: Improved modeling of thesequence-energy landscape of protein structures

This paper introduces PottsMPNN, a model trained to generate Potts energy functions for protein design, showing improved sequence generation and energy prediction over native sequence recovery.

How Not to be Seen: Predicting Unseen Enzyme Functions using Contrastive Learning

This paper introduces EnzPlacer, a contrastive learning method that predicts enzyme function from sequence by placing it in known function space, even when the precise function is unknown.