contrastive learning

Simultaneous dimensionality reduction and integration for single-cell ATAC-seq data using deep learning

This paper introduces a variational auto-encoder for single-cell ATAC-seq data that performs dimensionality reduction and batch correction simultaneously.

Automatic identification of intestinal parasites in reptiles using microscopic stool images and convolutional neural networks

The paper develops a CNN-based method to identify six intestinal parasites in reptiles using stool images, achieving 98.66% accuracy.

Framework for Denoising Monte Carlo Photon Transport Simulations Using Deep Learning

This paper develops a deep learning framework to denoise Monte Carlo photon transport simulations, significantly improving image quality and reducing computational costs.

Redefining the architecture of ferlin proteins: insights into multi-domain protein structure and function

The paper uses RoseTTAFold and AlphaFold2 to model human ferlin proteins, revealing a previously unknown C2 domain, C2-FerA, and consistent domain structures across ferlin proteins.

conST: an interpretable multi-modal contrastive learning framework for spatial transcriptomics

conST learns embeddings from gene expression, spatial neighbors, and histology via three-level contrastive learning, reaching adjusted Rand index 0.65 on spatialLIBD slice 151673, with GNNExplainer-based explanations.

Label2label: Training a neural network to selectively restore cellular structures in fluorescence microscopy

This paper introduces a neural network method to restore fluorescence microscopy images by reducing background signals from non-specific labeling.

Improved AlphaFold modeling with implicit experimental information

The paper improves AlphaFold modeling by incorporating experimental density maps iteratively, resulting in more accurate protein structure predictions.

Raman Signal Denoising using Fully Convolutional Encoder Decoder Network

This paper introduces a fully convolutional encoder-decoder network for Raman signal denoising, achieving better performance than traditional methods with reduced exposure times.

Conservative Significance Testing of Tripartite Interactions in Multivariate Neural Data

The paper evaluates tripartite interaction metrics in neural data, showing they are biased by impurities and proposes a conservative testing method to reduce false positives.

FAME: Fragment-based Conditional Molecular Generation for Phenotypic Drug Discovery

FAME generates novel molecules with desired biological activity by learning from gene expression data using a conditional variational autoencoder and fragment-based generation.

Natural gradient enables fast sampling in spiking neural networks

This paper shows how natural gradient methods can enable fast sampling in spiking neural networks by leveraging population geometry for probabilistic inference.

Attention-based deep multiple instances learning for classifying circular RNA and other long non-coding RNA

The paper introduces an attention-based deep multiple instance learning model for circRNA identification from lncRNA, showing improved performance and identifying key sequence motifs.

Modeling the Three-Dimensional Chromatin Structure from Hi-C Data with Transfer Learning

This paper introduces TECH-3D, a transfer learning method for 3D chromatin structure prediction from Hi-C data, outperforming previous methods without assuming contact frequency equals Euclidean distance.

Contrastive Cycle Adversarial Autoencoders for Single-cell Multi-omics Alignment and Integration

This paper introduces a framework to align and integrate single-cell RNA-seq and ATAC-seq data, achieving higher accuracy than existing methods on both simulated and real data.

Alzheimer's Disease Classification Accuracy is Improved by MRI Harmonization based on Attention-Guided Generative Adversarial Networks

This study improves Alzheimer's disease classification using MRI harmonization with an attention-guided GAN, showing better performance than non-harmonized images.

Molecular Graph Contrastive Learning with Parameterized Explainable Augmentations

This paper introduces MolCLE, a molecular graph contrastive learning framework with parameterized explainable augmentations, which outperforms baselines on real-world datasets with fewer parameters.

GraphPrompt: Biomedical Entity Normalization Using Graph-based Prompt Templates

This paper introduces GraphPrompt, a graph-based prompt learning method for biomedical entity normalization, improving performance on zero-shot and few-shot settings using a curated dataset of entity-synonym pairs.

Interactions of host defense and hyper-keratinization in psoriasis

The paper identifies a core gene network linked to inflammation and hyper-keratinization in psoriatic skin, associated with disease severity and microbiome composition.

Unexpected similarity between HIV-1 reverse transcriptase and tumor necrosis factor revealed by binding site image processing

The paper identifies an unexpected similarity between HIV-1 reverse transcriptase and tumor necrosis factor binding sites using a new computational method, confirmed by experimental assays.

Sampling the conformational landscapes of transporters and receptors with AlphaFold2

This study uses AlphaFold2 to generate multiple conformations of transporters and GPCRs, showing that reducing MSA depth improves conformational sampling accuracy.

Jointly Optimized Spatial Histogram UNET Architecture (JOSHUA) for Adipose Tissue Segmentation

This paper introduces JOSHUA, a CNN architecture with spatial histogram layers for adipose tissue segmentation in histological images, showing improved performance over baseline models.

CLMB: deep contrastive learning for robust metagenomic binning

CLMB improves metagenomic binning by using deep contrastive learning to handle noise, outperforming existing methods and recovering more high-quality genomes.

MapCell: Learning a comparative cell type distance metric with Siamese neural nets with applications towards cell-types identification across experimental datasets

MapCell uses Siamese neural networks to learn a cell type distance metric, enabling annotation transfer across scRNA-seq datasets and identifying novel cell types.

Detecting ice artefacts in processed macromolecular diffraction data with machine learning

The paper develops a machine learning tool called Helcaraxe to detect ice artefacts in processed macromolecular diffraction data, achieving better performance than previous methods.

CpG Transformer for imputation of single-cell methylomes

The paper introduces CpG Transformer, a model for imputing single-cell methylomes, achieving state-of-the-art performance on scBS-seq and scRRBS-seq datasets.

Generative modeling of histology tissue reduces human annotation effort for segmentation model development.

This paper uses datasetGAN to reduce annotation effort for glomeruli segmentation, showing it requires less annotation than traditional methods while maintaining high performance.

scNym: Semi-supervised adversarial neural networks for single cell classification

scNym transfers cell identity annotations across single-cell experiments using semi-supervised adversarial neural networks, outperforming existing methods.

Structures of core eukaryotic protein complexes

The paper uses coevolution analysis and deep learning to model core eukaryotic protein complexes in yeast, identifying 104 new assemblies and structurally characterizing 608 previously unknown complexes.

Generative Adversarial Networks (GAN) for the simulation of central-place foraging trajectories

This paper uses GAN to simulate central-place foraging trajectories, showing better performance than HMM in capturing large-scale statistical properties of animal movement.

Literature-Based Discovery beyond the ABC paradigm: a contrastive approach

This paper introduces a contrastive approach to Literature-Based Discovery (LBD) to improve relation identification between concepts in biomedical literature, demonstrating its effectiveness through case studies on neurodegenerative diseases.