David Fenyö

Proteogenomic analysis of up to 11 CPTAC cohorts shows ubiquitin-proteasome proteins often diverge from transcripts; TP53-mutant tumors elevate UBR5 protein, while TRIM28 prognosis is lineage-dependent.

This study identifies chromosome 1q gain as a biomarker for poor immunotherapy response in melanoma by linking aneuploidies to immune evasion mechanisms.

This paper introduces a method to deliver synthetic haplotypes containing T2D GWAS variants to hiPSCs to study their functional impact on differentiation.

The paper analyzes ORF1p expression across cancers, finding it correlates with immune suppression in some tumors and activation in others, linked to copy number alterations.

This paper introduces a segmentation-free method for analyzing multispectral whole slide images to study tissue repair and regeneration, revealing tissue compartments and dynamics without prior labels.

Chromnitron predicts protein-DNA binding landscapes across cell types and proteins using multimodal learning, revealing new regulators of T cell exhaustion and dynamic shifts during neurogenesis.

SimSpace simulates spatial omics data to enable method development and benchmarking, reproducing spatial statistics and supporting deconvolution and gene detection evaluations.

This paper introduces SGE, a method for screening metabolic pathways in mammalian cells, and finds that mitochondrial localization and large synthetic DNA constructs are key for successful pathway engineering.

This study integrates histopathology and proteogenomic data from melanoma metastases to identify subtypes and risk factors, offering a basis for personalized treatment strategies.

The paper introduces a segmentation-free vision transformer, CANVAS, to discover and characterize tumor microenvironment signatures from high-dimensional multiplexed images, identifying a monocytic signature linked to poor prognosis.

This paper identifies potential drug targets and predicts treatment efficacy and resistance in melanoma subtypes using proteogenomic data and drug sensitivity analysis.

This paper introduces a method to temporally control Cas9 DNA targeting by coupling it with transcription through engineered RNA binding, enabling selective gene editing in yeast and human cells.

The paper identifies six proteins that predict response to immune checkpoint therapy across three melanoma cohorts and correlate with patient survival.

This study shows D-type cyclins inhibit mismatch repair (MMR) in G1 phase by shielding p21, while their degradation in S phase enables proper MMR function, maintaining genome stability.

This paper develops a machine learning approach to classify 53BP1 foci in 2D images, achieving high performance with standard classifiers and a diverse parameter space.

This study identifies CD97 as a targetable vulnerability in glioblastoma through its role in tumor metabolism and signaling, supported by transcriptomic, proteomic, and functional analyses.

KaryoTap enables high-throughput single-cell aneuploidy profiling, revealing that positive and negative selection shape cancer genomes by influencing chromosomal gains and losses.

The paper develops a method to measure LINE-1 retrotransposon expression in single cells and identifies its presence in epithelial and neuronal cells, with age-related increases.

This study introduces C.Origami, a deep learning model that predicts cell type-specific 3D chromatin organization, enabling in silico genetic screening to identify genomic regions affecting chromatin structure.

MenDEL is a web tool that automatically designs PCR primers for DNA assemblies by optimizing primer pairs for each junction based on user constraints.

MenDEL automates the selection of BAC sets to cover long DNA regions by allowing users to define criteria and find optimal combinations from existing libraries.

This study uses a deep learning model to predict STK11 mutations in lung adenocarcinoma histopathology slides and finds that immune cell density is the key feature for classification.

This paper introduces a deep learning model that predicts endometrial cancer subtypes and mutations from histopathology images, offering a faster alternative to genomic sequencing.

The paper uses deep neural networks to predict gene perturbation from transcriptional profiles, achieving 74.9% accuracy but with limited generalizability due to data consistency issues.

BlackSheep identifies molecules aberrantly enriched in small sample cohorts using non-parametric differential extreme value analysis.

The paper introduces HotPho, a tool that identifies 3D co-clustering of phosphosites and cancer mutations on protein structures, revealing 474 hybrid clusters with potential functional significance.

An ensemble of six imputation methods, built through the NCI-CPTAC Proteogenomics DREAM Challenge, fills missing values in labeled LC-MS/MS proteomics data, helping most for low-abundance proteins.

The paper transplants the human adenine de novo pathway into yeast to study regulatory mechanisms, finding that PPAT/ADE4 is a central regulatory node influenced by metabolites and mutations.

The paper identifies functionally important regions in ORF1p and ORF2p of human LINE-1 through trialanine mutagenesis, showing most mutants lose retrotransposition activity.

L1EM quantifies LINE-1 RNA at specific genomic loci using expectation maximization, distinguishing retrotransposition-capable transcripts from non-functional ones.