sequence-to-function modeling

3-D structure based prediction of SUMOylation sites

The paper introduces a 3-D structure-based method to predict SUMOylation sites with 81% accuracy and a Matthews Correlation Coefficient of 0.4.

DBpred: A deep learning method for the prediction of DNA interacting residues in protein sequences

DBpred predicts DNA-interacting residues in proteins using deep learning, achieving improved AUROC scores compared to existing methods.

Interpretable RNA Foundation Model from Unannotated Data for Highly Accurate RNA Structure and Function Predictions

A BERT-style model pretrained on 23.7 million unlabelled ncRNA sequences gives embeddings that improve secondary structure, contact, distance-map and some function predictions over task-specific and MSA-based baselines.

Now What Sequence? Pre-trained Ensembles for Bayesian Optimization of Protein Sequences

The paper uses pre-trained sequence models with Bayesian optimization to design protein sequences with minimal labeled data, achieving efficient peptide inhibitor design.

Intrinsically disordered regions that drive phase separation form a robustly distinct protein class

The paper identifies amino acid property scales that distinguish phase-separating intrinsically disordered regions from other protein regions and improves a predictor for these regions.

TM-Vec: template modeling vectors for fast homology detection and alignment

TM-Vec predicts structural similarity from sequence pairs to detect remote homologs, outperforming sequence alignment methods and matching structure-based alignment performance.

Harnessing machine translation methods for sequence alignment

This paper introduces BetaAlign, a sequence alignment method using NLP techniques, which outperforms existing tools like MAFFT and MUSCLE.

Template-based assembly of proteomic short reads for de novo antibody sequencing and repertoire profiling

The paper introduces Stitch, a software tool for mapping proteomic short reads to templates to sequence monoclonal and profile polyclonal antibodies with high accuracy.

Efficient algorithms for designing maximally sized orthogonal DNA sequence libraries

The paper introduces SeqWalk, an efficient algorithm for designing large orthogonal DNA sequence libraries by minimizing sequence symmetry in a de Bruijn graph representation.

The DynaSig-ML Python package: automated learning of biomolecular dynamics-function relationships

The paper introduces DynaSig-ML, a Python package that learns dynamics-function relationships in biomolecules using sequence variants and the Elastic Network Contact Model.

Interpretable Pairwise Distillations for Generative Protein Sequence Models

The paper evaluates three neural network models for protein sequences, showing that pairwise models can replicate their performance in mutational effect prediction.

Direct Generation of Protein Conformational Ensembles via Machine Learning

This paper shows that machine learning can generate protein conformational ensembles without simulation, using a transformer-based GAN trained on simulation data.

Global, Highly Specific and Fast Filtering of Alignment Seeds

The paper introduces geometric hashing, a method to filter alignment seeds with high specificity and efficiency, reducing false positives in homology searches across human and mouse genomes.

Explainable deep graph learning accurately modeling the peptide secondary structure prediction

This paper introduces PHAT, a deep graph learning framework for peptide secondary structure prediction, achieving better accuracy and interpretability for short peptides.

Convolutions are competitive with transformers for protein sequence pretraining

This study shows that convolutional neural networks can be as effective as transformers for protein sequence pretraining, offering better scalability and performance on long sequences.

DCAlign v1.0: Aligning biological sequences using co-evolution models and informative priors

DCAlign v1.0 improves sequence alignment by incorporating co-evolution models and informative priors to reduce computational demands.

Alignment of Single-Molecule Sequencing Reads by Enhancing the Accuracy and Efficiency of Locality-Sensitive Hashing

The paper improves the accuracy and efficiency of MinHash for aligning single-molecule sequencing reads by using a single hash function and allowing one sequencing error in seed sequences, outperforming existing aligners on E. coli and human genomes.

Incorporating Target-Specific Pharmacophoric Information Into Deep Generative Models For Fragment Elaboration

STRIFE improves fragment elaboration by incorporating pharmacophoric information from protein targets, outperforming existing methods in generating ligand-efficient structures.

phastSim: efficient simulation of sequence evolution for pandemic-scale datasets

The paper introduces phastSim, a tool for efficiently simulating sequence evolution on large phylogenetic trees, particularly for SARS-CoV-2-like datasets.

Proximal Exploration for Model-guided Protein Sequence Design

This paper introduces Proximal Exploration (PEX) and MuFacNet to improve protein sequence design by prioritizing low-mutation high-fitness mutants and predicting mutational effects.

Model evolution in SARS-CoV-2 spike protein sequences using a generative neural network

The paper uses a generative neural network to model the evolution of SARS-CoV-2 spike protein sequences between adjacent clades, showing high correlation with true evolutionary patterns.

Drug design and repurposing with a sequence-to-drug paradigm

This paper introduces a sequence-to-drug paradigm using TransformerCPI2.0 to discover drug-like molecules from protein sequences and validates its effectiveness for three targets.

Seqrutinator: Non-Functional Homologue Sequence Scrutiny for the Generation of large Datatsets for Protein Superfamily Analysis

Seqrutinator identifies and removes non-functional homologue sequences from proteomes to improve multiple sequence alignments in plant superfamilies.

Generative and interpretable machine learning for aptamer design and analysis of in vitro sequence selection

This study uses Restricted Boltzmann Machines to analyze SELEX data for thrombin aptamers, enabling sequence fitness estimation, classification, and novel molecule generation, which is validated experimentally.

AbDiver - A tool to explore the natural antibody landscape to aid therapeutic design

AbDiver enables comparison of therapeutic antibody sequences to natural antibody diversity, aiding in rational design by identifying mutation patterns and relevant sequence matches.

EpiBERTope: a sequence-based pre-trained BERT model improves linear and structural epitope prediction by learning long-distance protein interactions effectively

EpiBERTope is a BERT-based model that predicts linear and structural epitopes from protein sequences, achieving high AUC scores and outperforming traditional models.

PepSeA: Peptide Sequence Alignment and Visualization Tools to Enable Lead Optimization

PepSeA provides tools for peptide sequence alignment and visualization to support lead optimization in drug discovery.

Characterisation and sequence mapping of large RNA and mRNA therapeutics using mass spectrometry

This study develops an automated workflow for rapid characterization and sequence mapping of large RNA and mRNA therapeutics using mass spectrometry.

De novo mass spectrometry peptide sequencing with a transformer model

Casanovo, a transformer encoder-decoder, translates unbinned tandem mass spectrum peaks into peptide sequences, beating DeepNovo and PointNovo on peptide-level precision across nine held-out species while running faster than both.

X-MOL: large-scale pre-training for molecular understanding and diverse molecular analysis

X-MOL uses large-scale pre-training on 1.1 billion molecules to improve molecular analysis tasks through fine-tuning, achieving state-of-the-art results in property prediction, reaction analysis, and molecule generation.