variational autoencoder

Characterizing chromatin folding coordinate and landscape with deep learning

The paper uses a variational autoencoder to analyze chromatin structure fluctuations and identifies a reaction coordinate for chromatin folding, revealing insights into TAD formation and folding energetics.

Natural representation of composite data with replicated autoencoders

This paper introduces a method using replicated autoencoders optimized for local entropy to infer compositional features from noisy data, showing improved performance on synthetic and protein sequence data.

Sequential compression of gene expression across dimensionalities and methods reveals no single best method or dimensionality

The paper evaluates multiple compression methods and dimensionalities on gene expression data, finding no single best approach but showing benefits from ensemble methods across different dimensions.

GRASP: a Bayesian network structure learning method using adaptive sequential Monte Carlo

GRASP is a Bayesian network structure learning method that uses adaptive sequential Monte Carlo to improve network discovery from genomic data.

Quantifying the tradeoff between sequencing depth and cell number in single-cell RNA-seq

The paper quantifies the tradeoff between sequencing depth and cell number in single-cell RNA-seq, finding that deeper sequencing per cell provides greater benefits than increasing cell numbers beyond 15,000 reads per cell.

Geometric purity, kinematic scaling and dynamic optimality in drawing movements beyond ellipses

The paper investigates drawing movements beyond ellipses, showing they follow 2/3 power law scaling, have quasi-pure curvature spectra, and exhibit minimum jerk dynamics, but not due to mechanical work minimization.

Nanopore base-calling from a perspective of instance segmentation

This paper introduces a base-calling method for nanopore sequencing by framing it as a one-dimensional segmentation task, achieving competitive accuracy compared to existing methods.

Joint Motion Estimation and Source Identification using Convective Regularisation with an Application to the Analysis of Laser Nanoablations

The paper develops a variational method for joint motion estimation and source identification in 1D image sequences, applied to analyze laser nanoablations in Drosophila embryos.

Dilated Saliency U-Net for White Matter Hyperintensities Segmentation using Irregularity Age Map

This paper introduces a Dilated Saliency U-Net for WMH segmentation using an irregularity age map, achieving better performance with fewer parameters and shorter training time.

Single-Cell Data Analysis Using MMD Variational Autoencoder for a More Informative Latent Representation

This study compares Vanilla VAE and MMD-VAE on single-cell datasets, finding MMD-VAE provides a more informative latent representation.

Variational Autoencoder: An Unsupervised Model for Modeling and Decoding fMRI Activity in Visual Cortex

This paper uses a variational autoencoder to model and decode fMRI activity in the visual cortex, finding lower prediction accuracy compared to CNNs but successful decoding of latent variables.

SISUA: SemI-SUpervised generative Autoencoder for single cell data

The paper introduces a semi-supervised generative model for single-cell data using CITE-seq protein counts to improve gene expression analysis.

Graph-Based Method for Anomaly Detection in Functional Brain Network using Variational Autoencoder

The paper proposes a graph-based method using a Variational Autoencoder to detect anomalies in functional brain networks by analyzing irregular graph structures.

Project Dhaka: Variational Autoencoder for Unmasking Tumor Heterogeneity from Single Cell Genomic Data

Project Dhaka uses a variational autoencoder to analyze single-cell genomic data, revealing tumor subpopulations and their evolutionary relationships that other methods miss.

Variational Infinite Heterogeneous Mixture Model for Semi-supervised Clustering of Heart Enhancers

The paper introduces a semi-supervised clustering model for identifying active enhancers in mouse heart tissues using heterogeneous genomic data.

Stochastic Variational Inference for Bayesian Phylogenetics: A Case of CAT Model

The paper develops a variational Bayesian method to accelerate phylogenetic inference in the CAT model, achieving accurate results with significantly reduced computational time.

Predicting the Evolution of White Matter Hyperintensities in Brain MRI using Generative Adversarial Networks and Irregularity Map

This paper introduces DEP-GAN, a GAN model that predicts WMH evolution in small vessel disease using irregularity maps and two discriminators, achieving 72.01% accuracy in predicting WMH evolution.

On variational solutions for whole brain serial-section histology using the computational anatomy random orbit model

This paper develops a variational framework for aligning histology stacks to an atlas, addressing curvature non-identifiability and enabling accurate reconstruction of whole-brain structures.

Generative modeling and latent space arithmetics predict single-cell perturbation response across cell types, studies and species

scGen models single-cell perturbation responses across cell types, studies, and species using variational autoencoders and latent space arithmetics, showing accurate out-of-sample predictions.

Ccube: A fast and robust method for estimating cancer cell fractions

Ccube estimates cancer cell fractions and subclonal composition from whole-genome sequencing data using a probabilistic framework and variational inference.

Thermodynamic integration for dynamic causal models

This paper introduces thermodynamic integration for dynamic causal modeling, showing it provides more robust log evidence estimates than variational Bayes in complex models.

Unsupervised deep learning with variational autoencoders applied to breast tumor genome-wide DNA methylation data with biologic feature extraction

This study applies variational autoencoders to breast tumor DNA methylation data to identify differential methylation patterns and regulatory regions.

DeepNose: Using artificial neural networks to represent the space of odorants

This paper trains an autoencoder to represent odorant chemical space and uses it to predict human olfactory percepts with accuracy comparable to traditional molecular descriptors.

Parameter tuning is a key part of dimensionality reduction via deep variational autoencoders for single cell RNA transcriptomics

The paper evaluates the Tybalt VAE for scRNA-seq data, showing its performance depends heavily on parameter tuning and outperforms other methods when properly tuned.

Autoencoder and Optimal Transport to Infer Single-Cell Trajectories of Biological Processes

This paper introduces ImageAEOT, a method combining autoencoders and optimal transport to infer single-cell trajectories from imaging data, validated on chromatin images of cancer cells and tissues.

VASC: dimension reduction and visualization of single cell RNA sequencing data by deep variational autoencoder

VASC is a deep variational autoencoder for scRNA-seq data that improves dimension reduction and visualization by modeling dropout events and nonlinear features.

Bayesian-estimated hierarchical HMMs enable robust analysis of single-molecule kinetic heterogeneity

This paper introduces hFRET, a computational tool using Bayesian hierarchical HMMs to analyze single-molecule kinetic heterogeneity, demonstrating its effectiveness with simulated and experimental data.

Predicting aging of brain metabolic topography using variational autoencoder

The paper predicts age-related brain metabolic changes using a variational autoencoder and shows APOE4 status significantly affects metabolic degeneration in specific brain regions.

Deep clustering of protein folding simulations

This paper uses a convolutional variational autoencoder to cluster protein folding simulations and identify conformational substates, achieving 89% contact prediction accuracy.

DeepProfile: Deep learning of cancer molecular profiles for precision medicine

DeepProfile trains a variational autoencoder on thousands of unlabeled gene expression samples to learn low-dimensional representations that improve prediction of acute myeloid leukemia drug responses and cancer phenotypes.