contrastive learning

S-PLM: Structure-aware Protein Language Model via Contrastive Learning between Sequence and Structure

S-PLM improves protein language models by integrating 3D structure information through contrastive learning, outperforming sequence-only models in clustering and classification tasks.

PathOmCLIP: Connecting tumor histology with spatial gene expression via locally enhanced contrastive learning of Pathology and Single-cell foundation model

PathOmCLIP infers spatial gene expression from histology images by combining pathology and single-cell RNA-seq foundation models, achieving better gene expression prediction than existing methods.

Evaluating generalizability of artificial intelligence models for molecular datasets

This paper evaluates how well deep learning models generalize to new molecular sequences using a new framework called SPECTRA across 18 datasets and 19 models.

inVAE: Conditionally invariant representation learning for generating multivariate single-cell reference maps

inVAE improves cell state identification by disentangling biological and technical variation in single-cell data, outperforming existing methods across multiple human tissues.

Cell Type Differentiation Using Network Clustering Algorithms

This study compares clustering algorithms for cell type annotation in scRNA-seq data, finding that Infomap outperforms others in identifying cell types in PBMC and ROSMAP datasets.

ACE: a versatile contrastive learning framework for single-cell mosaic integration

ACE provides a contrastive learning framework for mosaic integration of single-cell multi-omics data, improving representation of cellular heterogeneities in datasets with incomplete modalities.

Graph Contrastive Learning of Subcellular-resolution Spatial Transcriptomics Improves Cell Type Annotation and Reveals Critical Molecular Pathways

Focus improves cell type annotation by modeling subcellular RNA distribution in spatial transcriptomics data and identifies cell type-specific molecular pathways.

Quantitative Analysis of Miniature Synaptic Calcium Transients Using Positive Unlabeled Deep Learning

This paper introduces a PU learning-based method for detecting and segmenting miniature synaptic calcium transients, showing improved performance over traditional methods in cultured rat neurons.

Accurate structure prediction of immune proteins using parameter-efficient transfer learning

The paper introduces ImmuneFold, a transfer learning method for immune protein structure prediction, showing improved accuracy over existing methods for TCR-epitope binding.

FAPM: Functional Annotation of Proteins using Multi-Modal Models Beyond Structural Modeling

FAPM improves protein functional annotation by combining protein sequence and language models, outperforming sequence-only methods on benchmarks and phage proteins with few homologs.

Capturing cell heterogeneity in representations of cell populations for image-based profiling using contrastive learning

The paper introduces CytoSummaryNet, a contrastive learning method for aggregating single-cell features in image-based profiling, improving mechanism of action prediction by 30-68%.

Prediction of context-specific regulatory programs and pathways using interpretable deep learning

The paper introduces COBRA, an interpretable deep learning model for scRNA-seq data that disentangles covariate effects, and demonstrates its application on interferon-stimulated mouse datasets.

In vivo cell-type and brain region classification via multimodal contrastive learning

This paper introduces NEMO, a multimodal contrastive learning method for classifying cell types and brain regions from electrophysiological data, achieving state-of-the-art results on two opto-tagged datasets and the International Brain Laboratory Brain-wide Map dataset.

mcBERT: Patient-Level Single-cell Transcriptomics Data Representation

mcBERT generates patient-level representations from scRNA-seq data using a transformer-based model, enabling disease cohort identification and phenotype classification across diverse tissues.

Quantifying the intricacies of biological pattern formation: A new perspective through complexity measures

The paper introduces DNOS and DPC measures to quantify pattern diversity and complexity in nonlinear systems, demonstrating their application across multiple biological models.

A Multimodal Foundation Model for Discovering Genetic Associations with Brain Imaging Phenotypes

COMICAL discovers genetic associations with brain imaging phenotypes using contrastive learning and multi-omics data from the UK Biobank.

Cytometry Masked Autoencoder: An Accurate and Interpretable Automated Immunophenotyper

The paper introduces cyMAE, a masked autoencoder for automated immunophenotyping, showing it improves accuracy and interpretability over manual methods in cytometry data analysis.

MARVEL: Microenvironment Annotation by Supervised Graph Contrastive Learning

MARVEL improves microenvironment annotation by using supervised graph contrastive learning to embed spatial cell relationships into a continuous space, outperforming existing methods in three tasks.

Leveraging Lineage Barcodes as Natural Augmentations for Contrastive Learning of Cell Fate in scRNA-seq Data

This paper introduces LCL, a contrastive learning framework using lineage barcodes to isolate fate-determining signals in scRNA-seq data, showing improved performance over standard models.

Epitope-anchored contrastive transfer learning for paired CD8+ T cell receptor-antigen recognition

This paper introduces EPACT, a deep-learning framework for predicting TCR-antigen binding pairs, showing high accuracy in epitope recognition and TCR repertoire analysis.

Mapping lineage-traced cells across time points with moslin

The paper introduces moslin, a method that combines lineage and gene expression data across time points to predict cell fate and decision driver genes.

Du-IN-v2: Unleashing the Power of Vector Quantization for Decoding Cognitive States from Intracranial Neural Signals

The paper introduces Du-IN-v2, a pre-training framework for speech decoding from intracranial neural signals, achieving state-of-the-art performance using region-level token embeddings and discrete codex-guided mask modeling.

Multimodal learning of transcriptomes and text enables interactive single-cell RNA-seq data exploration with natural-language chats

CellWhisperer enables chat-based exploration of single-cell RNA-seq data using natural language by connecting transcriptomes with text through multimodal learning.

Exploring Protein-DNA Binding Residue Prediction and Consistent Interpretability Analysis Using Deep Learning

This study introduces a sequence-based deep learning method for predicting DNA-binding residues, outperforming existing models and demonstrating interpretability through contrastive learning and graph neural networks.

Single-Cell Data Integration and Cell Type Annotation through Contrastive Adversarial Open-set Domain Adaptation

SAFAARI integrates single-cell data, corrects batch effects, and annotates cell types using contrastive adversarial domain adaptation, outperforming existing methods in multiple benchmarks.

A 3D Image Segmentation Study of U-Net on CT Images of the Human Aorta with Morphologically Diverse Anatomy

This study compares 2D and 3D U-Net for segmenting human aorta from CT images, finding 3D U-Net performs better but has geometry reconstruction errors.

Unlearning Virus Knowledge Toward Safe and Responsible Mutation Effect Predictions

This paper introduces PROEDIT, a method to remove virus-related knowledge from pre-trained protein models to improve safety in mutation effect predictions without affecting performance on non-virus proteins.

EmbedSimScore: Advancing Protein Similarity Analysis with Structural and Contextual Embeddings

EmbedSimScore improves protein similarity analysis by integrating structural and contextual embeddings to capture both local and global features.

Contrastive learning explains the emergence and function of visual category-selective regions

This paper introduces contrastive coding as a framework explaining category-selective regions in the human brain using neural network analyses.

Interpreting Sleep Activity Through Neural Contrastive Learning

The paper develops a neural network model, Sleep Interpreter (SI), to decode memory replay from sleep EEG and shows it can accurately detect reactivation during sleep that correlates with memory performance.