sequence-to-function modeling

SeqForge: A scalable platform for alignment-based searches, motif detection, and sequence curation across meta/genomic datasets

SeqForge provides a scalable toolkit for alignment-based searches, motif detection, and sequence curation in metagenomic datasets, reducing the need for custom scripting.

Hit or Miss: Understanding Emergence and Absence of Homo-oligomeric Contacts in Protein Language Models

This study shows that large protein language models can predict homo-oligomeric interface contacts, with performance improving with model size and alignment diversity.

SignalGen: A Protein Language Model Based AI Agent For Optimal Signal Peptide Prediction

SignalGen predicts optimal signal peptides for protein expression by integrating mature protein sequences, organism, and localization using a Latent Residual Transformer.

Overcoming Extrapolation Challenges of Deep Learning by Incorporating Physics in Protein Sequence-Function Modeling

This paper improves deep learning models for protein sequence-function prediction by integrating physical principles to enhance extrapolation performance.

Large mRNA language foundation modeling with NUWA for unified sequence perception and generation

This paper introduces NUWA, an mRNA language model that enables unified sequence perception and generation across diverse species and tasks.

A Genetic Algorithm to scour protein sequence space: A novel framework for protein engineering using protein language models and force fields

GAPO uses genetic algorithms with protein language models and force fields to optimize protein sequences, outperforming simulated annealing in binding and stability.

MillionFull enables massive, full-length enzyme sequence-fitness data collection at low cost for machine learning-guided enzyme engineering

MillionFull collects large-scale sequence-fitness data for enzymes at low cost, enabling machine learning for enzyme engineering.

Efficient trace reconstruction in DNA storage systems using Bidirectional Beam Search

This paper introduces Bidirectional Beam Search for efficient trace reconstruction in DNA storage systems, achieving high accuracy and speed on real datasets.

Decoding the physicochemical basis of taxonomy preferences in protein design models

This study examines how scoring context and training data influence protein design models' preferences across species, showing that backbone-conditioned models favor archaeal and thermophilic proteins while sequence-only models favor eukaryotic proteins.

All-atom protein design via SE(3) flow matching with ProteinZen

This paper introduces ProteinZen, a method for all-atom protein design using SE(3) flow matching, achieving high sequence-structure consistency and competitive performance in design tasks.

PepGlider: Property Regularized VAE for interpretable and controllable peptide design

PepGlider enables precise control over peptide properties through a continuous regularization framework, allowing for de novo generation and targeted optimization of peptides with desired characteristics.

ProT-VAE: Protein Transformer Variational AutoEncoder for Functional Protein Design

ProT-VAE designs functional proteins by learning sequence-function relationships using a transformer-based variational autoencoder, enabling conditional generation of novel sequences with high functionality.

Generalized design of sequence-ensemble-function relationships for intrinsically disordered proteins

This paper introduces a computational framework for designing intrinsically disordered proteins by inverting molecular simulations to create sequence-ensemble-function relationships.

RNAtranslator: Modeling protein-conditional RNA design as sequence-to-sequence natural language translation

RNAtranslator generates RNA sequences that bind to specific proteins by treating RNA design as a sequence-to-sequence translation task, producing novel and high-affinity sequences.

RNA sequence design and protein-DNA specificity prediction with NA-MPNN

This paper introduces NA-MPNN, a unified deep learning model for nucleic acid inverse folding, which outperforms existing methods in RNA sequence design and protein-DNA specificity prediction.

Split inteins for generating combinatorial non-ribosomal peptide libraries

This paper introduces split inteins to enable modular NRPS engineering, generating 324 hybrid NRPS and over 200 novel peptides through combinatorial transformation.

Pretrained protein language models choose between sequence novelty and structural completeness

This paper shows that protein language models prioritize sequence novelty over structural diversity, leading to a tradeoff in generating realistic protein structures.

Isolation of structural constraints by ensemble-conditioned protein sequence design

This paper introduces Caliby, a sequence design method that conditions on structural ensembles to improve designability while reducing native sequence recovery.

Design of proteins by parallel tempering in the sequence space

The paper designs proteins using parallel tempering in sequence space to improve structure prediction and globularity.

Yorzoi: Predicting RNA-seq coverage from DNA sequence in yeast

Yorzoi predicts RNA-seq coverage for yeast DNA sequences, enabling in-silico testing and design of synthetic DNA constructs.

GPU-accelerated homology search with MMseqs2

The paper introduces GPU-accelerated homology search with MMseqs2, achieving faster sequence similarity detection than CPU methods and improving structure prediction speeds.

Seasonal influenza viruses show distinct adaptive dynamics during growth in chicken eggs

The paper examines adaptive mutations in seasonal influenza viruses grown in chicken eggs, finding most mutations cluster around the HA receptor-binding pocket and show varying background dependence across subtypes.

Painting Peptides with Antimicrobial Potency through Deep Reinforcement Learning

This paper introduces AMPainter, a deep reinforcement learning model for designing antimicrobial peptides, which outperforms existing models and generates effective AMPs from signal peptides and random sequences.

NucleicBERT: Deciphering the language of nucleic acids

NucleicBERT is a self-supervised masked-language model for RNA that learns contextual representations without alignments or evolutionary data, outperforming existing models in structural and functional tasks.

Uncovering differential tolerance to deletions versus substitutions with a protein language model

The paper uses a protein language model to analyze amino acid deletions in the human proteome, identifying sites with differential tolerance to deletions versus substitutions.

Deep generative models generate mRNA sequences with enhanced translation capacity and stability

GEMORNA generates mRNA sequences with enhanced translation capacity and stability, showing significant improvements in expression levels and therapeutic applications.

Fast Multi-objective RNA Optimization with Autoregressive Reinforcement Learning

RNAJog optimizes RNA sequences for multiple objectives using autoregressive reinforcement learning, achieving faster performance and improved vaccine design outcomes.

Deep generative modeling reveals maturation-linked pairing signatures in human antibodies

This paper develops a deep learning framework to generate light chains from heavy chains using unpaired antibody sequences, revealing maturation-linked pairing patterns.

ProtRNA: A Protein-derived RNA Language Model by Cross-Modality Transfer Learning

ProtRNA transfers knowledge from protein language models to RNA using cross-modality transfer learning, achieving comparable or better performance with less data and parameters.

Correlated Segments of Intrinsically Disordered Proteins as Drivers of Homotypic Phase Separation

The paper identifies correlated segments in intrinsically disordered proteins as key drivers of homotypic phase separation, showing that these segments, not individual residues, are responsible for the phenomenon.