Ziv Bar‐Joseph

The paper finetunes gene embeddings from foundation models for temporal clinical transcriptomics data, improving gene network analysis and detection of disease progression and drug response mechanisms.

This study evaluates foundation models for predicting cellular responses to perturbations, finding that some pretrained embeddings improve predictions while others do not.

This paper introduces DeepSAS, a deep graph framework to identify senescent cells and their associated genes, revealing cell-type-specific senescence features in aging and disease.

This paper introduces GIANT, a method for integrating atlas-scale single cell data from multiple modalities and tissues, creating a unified gene embedding space that reflects gene functions and regulations.

PyEvoCell uses an LLM to aid in interpreting single-cell trajectory analysis by suggesting lineages, performing differential expression analysis, and validating hypotheses with PubMed citations.

ENACT improves cell type annotation and transcript mapping in Visium HD spatial transcriptomics data through a tissue-agnostic pipeline.

The paper applies a novel active learning strategy, DEW-DROP, to improve protein structure prediction efficiency and data selection for two protein datasets.

The paper develops SenSet, a gene signature for human lung senescence, validated in 3D culture models and identifies cell-type specific senescence mechanisms in aging and disease.

The paper introduces Xron, a semi-supervised basecaller for detecting m6A methylation from nanopore sequencing by combining synthetic and experimental data.

CodonBERT is a large language model for mRNA design and optimization that outperforms previous methods on a flu vaccine dataset.

scResolve recovers single-cell expression profiles from multi-cellular spatial transcriptomics data, enabling cell type-specific analysis and rare population detection in human tissues.

The paper introduces iDiscoid, a human stem cell model of post-implantation development that recapitulates key morphogenetic features and offers a platform for studying early human development.

The paper develops two novel active learning batch selection methods for drug discovery, showing significant improvements over existing methods across multiple datasets.

SCS improves cell segmentation in sub-cellular spatial transcriptomics by combining imaging and sequencing data, achieving higher accuracy and better cell size estimation.

CINS infers differential cell-cell interactions between conditions using scRNA-Seq data and identifies the proteins involved in these interactions.

UNIFAN improves cell type annotation in scRNA-Seq by simultaneously clustering and annotating cells using gene sets, outperforming prior methods.

The paper introduces TraSig, a method that uses pseudotime ordering from scRNA-Seq data to improve the inference of cell-cell interactions by identifying ligand-receptor pairs with similar trajectories.

The paper introduces a computational pipeline, PALM, that integrates multi-omics time-series microbiome data using dynamic Bayesian networks to identify microbial and host interactions.

This paper develops deep learning methods to predict gene interactions from single-cell time-course RNA-Seq data, showing improved accuracy over existing methods.

Cellar is an interactive tool for annotating cell types in single-cell multi-omics and spatial proteomics data using HuBMAP datasets.

The paper introduces FICT, a method that improves cell type assignment in spatial transcriptomics data by incorporating both gene expression and neighborhood information, showing better accuracy in identifying cell types and sub-types.

This paper develops a domain adversarial neural network to reduce scRNA-seq data dimensionality while minimizing batch effects, improving cell type separation across datasets.

The paper introduces SCIPR, a method for aligning scRNA-seq data across studies, showing it can generalize to new data and identify cell type-specific genes.

MESSI identifies active signaling genes in spatial single-cell data by using a mixture of experts approach, improving prediction accuracy and providing biological insights.

This paper introduces LinTIMaT, a method that integrates CRISPR-Cas9 mutation and transcriptomic data to reconstruct lineage trees with improved accuracy and coherence.

The paper reconstructs signaling and regulatory networks for SARS-CoV-2 infection and identifies potential drug targets by integrating multi-omics data and experimental validation.

GCNG uses graph convolutional networks to infer cell-cell gene interactions from spatial transcriptomics data, improving upon existing methods and enabling downstream functional analysis.

The paper introduces a deep learning framework, CNNC, for inferring gene relationships from single-cell expression data, showing improvements over existing methods in various tasks.

The paper maps cell fate trajectories of human PSC-derived distal lung progenitors using scRNA-seq and identifies developmental plasticity regulated by Wnt signaling.

The paper introduces a continuous state HMM method for modeling time series scRNA-Seq data, showing improved accuracy in inferring branching topology and cell path assignments.