sequence-to-function modeling

A Framework for Accurate Prediction of Plastic-Degrading Enzymes using Convolutional Neural Networks

This paper develops a CNN-based framework to predict plastic-degrading enzymes from environmental sequences, achieving high accuracy and F1 score.

PARMIK: PArtial Read Matching with Inexpensive K-mers

PARMIK improves metagenomic sequence search by using inexpensive k-mers to achieve higher recall and faster alignment than BLAST.

Model-directed generation of CRISPR-Cas13a guide RNAs designs artificial sequences that improve nucleic acid detection

This paper designs artificial CRISPR-Cas13a guide RNAs using model-directed algorithms, showing improved sensitivity and variant discrimination in nucleic acid detection.

MethylSeqLogo: DNA methylation smart sequence logos

The paper introduces MethylSeqLogo, a method that integrates DNA methylation data into sequence logos to better represent transcription factor binding specificity.

ML-driven design of 3' UTRs for mRNA stability

The paper designs stable mRNA 3' UTRs using machine learning and experimental validation, achieving higher protein production in mouse models.

IgBlend: Unifying 3D Structures and Sequences in Antibody Language Models

IgBlend integrates 3D structures and sequences for antibody language models, outperforming existing models in benchmarks and showing correlation with binding affinities.

Recurrent Neural Network-based Prediction of O-GlcNAcylation Sites in Mammalian Proteins

The paper develops RNN and MLP models to predict O-GlcNAcylation sites in mammalian proteins, achieving significantly better performance metrics than previous models.

ProT-Diff: A Modularized and Efficient Approach to De Novo Generation of Antimicrobial Peptide Sequences through Integration of Protein Language Model and Diffusion Model

ProT-Diff generates novel antimicrobial peptides using a protein language model and diffusion model, producing 34 active peptides against bacteria with low toxicity.

RNA-DCGen: Dual Constrained RNA Sequence Generation with LLM-Attack

RNA-DCGen generates RNA sequences with specific structural or functional properties using a language model, outperforming random baselines in structure prediction.

An interpretable neural network unveils higher-order epistasis in large protein sequence-function relationships

This study introduces an interpretable neural network to analyze higher-order epistasis in protein sequence-function relationships, finding that higher-order interactions contribute up to 60% of the epistatic variance.

Manufacturing-Aware Generative Model Architectures Enable Biological Sequence Design and Synthesis at Petascale

The paper introduces manufacturing-aware generative models to enable efficient biological sequence design and synthesis at petascale, reducing costs by a trillion-fold.

The simplicity of protein sequence-function relationships

Reference-free analysis with a sigmoid link shows additive plus pairwise effects explain a median 96% of phenotypic variance across 20 combinatorial protein mutagenesis datasets.

Deep Learning Predicts Non-Normal Peptide FAIMS Mobility Distributions Directly from Sequence

The paper develops deep learning models to predict FAIMS ion mobility of peptides from sequence, achieving high F2 score on E. coli peptides.

Memory-bound k-mer selection for large evolutionary diverse reference libraries

This paper introduces KRANK, a k-mer selection algorithm that reduces memory usage while maintaining classification accuracy in taxonomic profiling.

Protein Language Models in Directed Evolution

This paper demonstrates how protein language models can guide directed evolution to improve PET degradation enzyme variants.

Multi-Modal Large Language Model Enables Protein Function Prediction

ProteinChat is a multi-modal large language model that predicts protein functions from amino acid sequences, outperforming general-purpose LLMs and task-specific models.

Comprehensive assembly of monoclonal and mixed antibody sequences

This paper introduces Fusion, a de novo assembler for antibody sequences, demonstrating high accuracy in reconstructing monoclonal and mixed antibody sequences from mass spectrometry data.

Toward De Novo Protein Design from Natural Language

This paper introduces Pinal, a foundation model that designs proteins from natural language descriptions, achieving functional alignment and experimental success in four design tasks.

A simple theory for finding related sequences by adding probabilities of alternative alignments

The paper introduces a simple modification to sequence alignment that sums probabilities of alternative alignments, improving detection of distant sequence relationships.

Sequence-to-sequence translation from mass spectra to peptides with a transformer model

Casanovo translates tandem mass spectra into peptide sequences using a transformer model, outperforming existing methods and improving analysis in immunopeptidomics and metaproteomics.

Generating information-dense promoter sequences with optimal string packing

The paper introduces a method to optimally pack DNA-protein binding sites into dense nucleotide sequences, solving an NP-hard problem using integer linear programming.

Transformer Graph Variational Autoencoder for Generative Molecular Design

This paper introduces TGVAE, a Transformer Graph Variational Autoencoder that generates diverse and chemically valid molecules by using molecular graphs instead of SMILES strings, outperforming existing methods.

CodonBERT: Large Language Models for mRNA Design and Optimization

CodonBERT is a large language model for mRNA design and optimization that outperforms previous methods on a flu vaccine dataset.

Simultaneous polyclonal antibody sequencing and epitope mapping by cryo electron microscopy and mass spectrometry - a perspective.

The paper evaluates using cryoEM and mass spectrometry with a deep-learning tool to sequence antibodies and map epitopes from convalescent serum, showing 80-90% accuracy in variable domain sequencing.

Optimizing 5'UTRs for mRNA-delivered gene editing using deep learning

The paper uses deep learning to design optimized 5'UTRs for mRNA-delivered gene editing, showing improved editing efficiency across cell types and targets.

Knot or Not? Sequence-Based Identification of Knotted Proteins With Machine Learning

This study uses machine learning to predict protein knotting from amino acid sequences, achieving 98.5% accuracy using AlphaFold predictions and UniProt data.

COATi: statistical pairwise alignment of protein-coding sequences

COATi improves sequence alignment accuracy by handling artifacts in protein-coding sequences, reducing data discarding in comparative genomic studies.

TumFlow: An AI Model for Predicting New Anticancer Molecules

TumFlow generates and optimizes new anticancer molecules using AI, showing improved synthetic feasibility and predicted efficacy against melanoma.

Protein Design with StructureGPT: a Deep Learning Model for Protein Structure-to-Sequence Translation

StructureGPT translates protein tertiary structures into amino acid sequences, improving solubility and stability for biomedical applications.

ProtEx: A Retrieval-Augmented Approach for Protein Function Prediction

ProtEx improves protein function prediction by using retrieved exemplars, achieving state-of-the-art results for EC numbers, GO terms, and Pfam families.