variational autoencoder

System-level health profiling from blood DNA methylation with explainable deep learning

The paper develops an explainable deep learning framework to predict health scores from blood DNA methylation, showing strong predictive accuracy and alignment with biological processes.

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.

Counting-based inference of mutant growth rates from pooled sequencing across growth regimes

The paper develops methods to infer mutant growth rates from pooled sequencing data by modeling growth and noise, improving accuracy over existing approaches.

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.

Optimal transport modeling uncovers spatial domain dynamics in spatiotemporal transcriptomics studies

SpaDOT identifies spatial domains and their temporal dynamics in spatiotemporal transcriptomics data, revealing key developmental processes like valvulogenesis.

Unsupervised learning of multi-omics data enables disease risk prediction in the UK Biobank

The paper applies a variational autoencoder to multi-omics data from the UK Biobank to assess disease risk prediction and the impact of preprocessing on model performance.

GCP-VQVAE: A Geometry-Complete Language for Protein 3D Structure

This paper introduces GCP-VQVAE, a geometry-complete tokenizer for protein structures, achieving state-of-the-art performance on benchmark datasets and demonstrating robust generalization to unseen proteins.

GRPa-PRS: A risk stratification method to identify genetically-regulated pathways in polygenic diseases

GRPa-PRS identifies genetically-regulated pathways in polygenic diseases by stratifying individuals based on PRS and analyzing differential gene expression across risk strata.

Merging conformational landscapes in a single consensus space with FlexConsensus algorithm

The paper introduces FlexConsensus, an algorithm that merges conformational landscapes from different CryoEM heterogeneity methods into a shared consensus space to improve reliability and reproducibility.

High-Fidelity Neural Speech Reconstruction through an Efficient Acoustic-Linguistic Dual-Pathway Framework

This paper introduces a dual-path framework for reconstructing high-fidelity speech from neural recordings, achieving high intelligibility with limited ECoG data.

Efficient protein structure generation with sparse denoising models

This paper introduces salad, a sparse denoising model for efficient protein backbone generation, and demonstrates its effectiveness for various protein design tasks.

Representation learning of single-cell time-series with deep variational autoencoders

This paper uses deep variational autoencoders to learn representations of single-cell time-series data from E. coli, showing good generalization across growth conditions and drug treatments.

Discovering cell types and states from reference atlases with heterogeneous single-cell ATAC-seq features

EpiPack improves scATAC-seq analysis by integrating heterogeneous features and detecting perturbed cell types and states through transfer learning and graph-based modeling.

Algebraic and Generative Design of DNA Origami via Origami Monoids

This work combines monoid structures with deep learning to design and analyze DNA origami, enabling systematic exploration of structural space.

Few-shot learning for highly accelerated 3D time-of-flight MRA reconstruction

This paper develops a few-shot learning method for 3D TOF-MRA reconstruction, achieving high-quality results with minimal experimental data and outperforming existing methods.

LieOTDock: A Protein Docking Framework via Lie Group Optimization and Optimal Transport

LieOTDock generates high-quality protein docking poses using Lie group optimization and optimal transport, achieving 3.43 Å RMSD on a benchmark in under three minutes.

PILOT-GM-VAE: Patient-Level Analysis of single cell Disease Atlas with Optimal Transport of Gaussian Mixture Variational Autoencoders

PILOT-GM-VAE analyzes single cell disease atlases to perform sample-level clustering, trajectory inference, and batch correction using Gaussian Mixture Variational Autoencoders with optimal transport.

DeepBioSim: Efficient and Versatile Methods for Microbiome Data Simulation with Minimal Statistical Assumptions

DeepBioSim generates realistic microbiome datasets using deep learning, without strong assumptions or phylogenetic input, and can also simulate single-organism omics profiles.

LieOTAlign: A Differentiable Protein Structure Alignment Framework Combining Optimal Transport and Lie Algebra

LieOTAlign introduces a differentiable protein structure alignment framework using Lie algebra and optimal transport, achieving higher TM-scores on benchmark datasets.

Neuronal Decoding of Decisions in Multidimensional Feature Space Using a Gated Recurrent Variational Autoencoder

This paper develops a gated recurrent variational autoencoder to decode decision-making from neuronal data in monkeys, showing comparable accuracy in multidimensional feature space.

GENESIS: Generating scRNA-Seq data from Multiome Gene Expression

GENESIS converts Multiome GEX data into synthetic scRNA-Seq profiles using generative models, improving cell-type classification accuracy.

Employing Artificial Neural Networks for Optimal Storage and Facile Sharing of Molecular Dynamics Simulation Trajectories

This paper develops a method using AutoEncoders to compress Molecular Dynamics trajectories, achieving 98% space reduction while maintaining accurate reconstruction of atom coordinates.

A Multi-Property Optimizing Generative Adversarial Network for de novo Antimicrobial Peptide Design

This paper introduces MPOGAN, a generative adversarial network for designing antimicrobial peptides with multiple desired properties, showing improved performance in antimicrobial activity and reduced cytotoxicity.

Deep Generative Models for Discrete Genotype Simulation

This study evaluates deep generative models for simulating discrete genotype data, showing their ability to capture genetic patterns and preserve genotype-phenotype associations.

Variational Autoencoder-based Model Improves Polygenic Prediction in Blood Cell Traits

This study introduces VAE-PRS, a deep learning model that improves polygenic prediction for blood cell traits by capturing interaction effects and outperforming traditional methods in 14 out of 16 traits.

Quantome: A Quantum Surrogate Model for Biophysical Landscapes

The paper introduces a quantum surrogate model framework for biophysical landscapes, demonstrating its ability to simulate transitions between localized and delocalized states using variational quantum eigensolver simulations.

Predicting individual learning trajectories in zebrafish via the free-energy principle

This paper uses active inference to model zebrafish neuronal dynamics during a go/no-go task, predicting individual learning trajectories and behavioural changes.

Interpretable spatial multi-omics data integration and dimension reduction with SpaMV

The paper introduces SpaMV, a method for integrating spatial multi-omics data that preserves modality-specific information and improves clustering and dimensionality reduction.

Triple-effect correction for Cell Painting data with contrastive and domain-adversarial learning

The paper introduces cpDistiller, a method to correct three technical effects in Cell Painting data using contrastive and domain-adversarial learning, improving analysis reliability and phenotypic response detection.

scPDA: Denoising Protein Expression in Droplet-Based Single-Cell Data

scPDA denoises protein expression in single-cell droplet data, improving cell-type identification by removing noise more thoroughly while preserving biological signals.