contrastive learning

Interpretable neural networks prioritize cancer driver genes from genome-wide dependency landscapes

The paper introduces two interpretable neural networks to prioritize cancer driver genes by analyzing genome-wide dependency landscapes and drug-response patterns.

Focused learning by antibody language models using preferential masking of non-templated regions

This paper introduces a preferential masking strategy for antibody language models to improve prediction of non-templated residues in CDR3 regions, enhancing classification and specificity.

Hierarchical cross-entropy loss improves atlas-scale single-cell annotation models

This paper introduces a hierarchical cross-entropy loss to improve single-cell annotation models by aligning them with biological hierarchies, achieving better out-of-distribution performance.

Detecting cell-level transcriptomic changes of Perturb-seq using Contrastive Fine-tuning of Single-Cell Foundation Models

The paper introduces a contrastive fine-tuning method for single-cell foundation models to detect subtle transcriptomic changes in Perturb-seq data, outperforming traditional differential expression methods.

Wavefront estimation through structured detection in laser scanning microscopy

This paper introduces a convolutional neural network for aberration correction in laser scanning microscopy using detector array data, achieving high-resolution imaging with minimal sample stress.

Self-contrastive learning enables interference-resilient and generalizable fluorescence microscopy signal detection without interference modeling

This paper introduces DEPAF, a self-contrastive learning method for fluorescence microscopy signal detection that is interference-resilient and generalizable without requiring interference modeling.

scContrast: A contrastive learning based approach for encoding single-cell gene expression data

scContrast is a contrastive learning method for embedding scRNA-seq data into a universal space, outperforming existing methods in integrating data from different sequencing protocols.

Dimensionality Reduction of Genetic Data using Contrastive Learning

This paper introduces a contrastive learning framework for dimensionality reduction of genetic data, showing improved preservation of global structure and generalization over existing methods.

SCARF: Single Cell ATAC-seq and RNA-seq Foundation model

SCARF is a foundation model for integrating single-cell ATAC-seq and RNA-seq data, achieving state-of-the-art performance in cell representation and cross-omics tasks.

Decoding the Moving Mind: Multi-Subject fMRI-to-Video Retrieval with MLLM Semantic Grounding

This paper introduces a framework for multi-subject fMRI-to-video retrieval that improves retrieval accuracy by addressing temporal mismatches, integrating subject-specific patterns, and using semantic grounding from LLMs.

CLCluster: a redundancy-reduction contrastive learning-based clustering method of cancer subtype based on multi-omics data

CLCluster uses contrastive learning with multi-omics data to cluster cancer subtypes, showing improved performance over existing methods and identifying AS events linked to patient survival.

Interpretable deep learning for deconvolutional analysis of neural signals

This paper introduces an interpretable deep learning method for deconvolving neural signals, demonstrating its application in analyzing brain activity across different regions and recording modalities.

Unsupervised Whole-Genome Representation Learning Captures Bacterial Phenotypes

This paper introduces a self-supervised whole-genome representation learning method for bacteria, showing it outperforms traditional feature-based approaches in predicting phenotypes.

Generative World Models to compute protein folding pathways

This paper develops a generative World Model to simulate protein folding pathways using dihedral rotations and validates it against MD data.

Enhancing Brain Age Prediction and Neurodegeneration Detection with Contrastive Learning on Regional Biomechanical Properties

This paper uses contrastive learning on regional biomechanical properties from MRE to improve brain age prediction and detect neurodegeneration, finding that stiffness and damping ratio are sensitive biomarkers for Alzheimer's and MCI.

A hybrid machine learning framework leveraging biophysicochemical insights for scalable discovery of protein-ligand interactions

The paper introduces a hybrid machine learning framework that improves protein-ligand interaction prediction by combining biophysicochemical insights with large-scale representations, achieving faster and more accurate virtual screening.

MuCST: restoring and integrating heterogeneous morphology images and spatial transcriptomics data with contrastive learning

MuCST integrates heterogeneous morphology images and spatial transcriptomics data using contrastive learning, identifying tissue subpopulations and tumor domains with high accuracy.

AGP-Net: A Universal Network for Gene Expression Prediction of Spatial Transcriptomics

AGP-Net predicts spatial gene expression from histopathology images by leveraging textual gene descriptions to generalize across unseen genes and datasets.

Self-Supervised Learning Reveals Clinically Relevant Histomorphological Patterns for Therapeutic Strategies in Colon Cancer

This paper uses self-supervised learning to identify histomorphological patterns in colon cancer WSIs, revealing clinically relevant clusters associated with survival and treatment responses.

Fine-Tuned Deep Transfer Learning Models for Large Screenings of Safer Drugs Targeting Class A G Protein-Coupled Receptors

The paper develops fine-tuned deep learning models for predicting safer drugs targeting class A GPCRs by leveraging transfer learning and sequence-based analysis.

Harnessing Generative Pre-trained Transformer for Antimicrobial Peptide Generation and MIC Prediction with Contrastive Learning

This paper introduces AMPCLGPT, a model that generates antimicrobial peptides and predicts their MIC values using contrastive learning and pre-training.

FusionESP: Improved enzyme-substrate pair prediction by fusing protein and chemical knowledge

FusionESP improves enzyme-substrate pair prediction by integrating protein and chemical language models with contrastive learning, achieving high accuracy with fewer resources.

Weakly Supervised Vector Quantization for Whole Slide Image Classification

This paper introduces VQ-MIL, a weakly supervised vector quantization method for WSI classification, achieving state-of-the-art results on benchmark datasets.

Predicting PROTAC-targeted Degradation and Designing Androgen Receptor Degraders with AiPROTAC

AiPROTAC predicts PROTAC degradation potential and designs an androgen receptor degrader, outperforming existing models and showing improved in vitro efficacy.

Reliable machine learning models in genomic medicine using conformal prediction

This study explores the use of conformal prediction in genomic medicine to improve the reliability and safety of machine learning models in clinical applications.

CyclicCAE: A Conformational Autoencoder for Efficient Heterochiral Macrocyclic Conformational Sampling

CyclicCAE generates energetically favorable macrocycle conformations faster than Rosetta's GeneralizedKIC, using synthetic data and enabling design and structure prediction.

Tissue Reconstruction with Generative AI

LUNA reconstructs tissue architecture from dissociated single-cell RNA-seq data using spatial priors learned from spatial transcriptomics, validating on mouse brain datasets and disease models.

Predicting mutation-disease associations through protein interactions via deep learning

This paper constructs a mutation-disease dataset and proposes CGM to predict mutation-disease associations, identifying 22 unknown protein interaction pairs.

MEDiCINe: Motion Correction for Neural Electrophysiology Recordings

MEDiCINe improves motion correction in neural electrophysiology recordings, leading to more accurate spike sorting.

Incorporating Hierarchical Information into Multiple Instance Learning for Patient Phenotype Prediction with scRNA-seq Data

This paper improves patient phenotype prediction using scRNA-seq data by incorporating hierarchical cell type information into multiple instance learning, achieving better performance and interpretability.