Olivier Elemento

This paper introduces UTAG, a method to automatically identify tissue structures in multiplexed imaging data, showing it can detect higher-order tissue organization and relate it to disease states.

This study shows that APOBEC3G drives mutagenesis and clonal evolution in bladder cancer, contributing to mutational profiles in multiple cancer types.

The paper introduces the GA4GH Phenopacket schema to standardize the exchange of phenotypic data for precision medicine applications.

This paper introduces a weakly-supervised method to predict tumor purity from H&E stained slides, achieving high accuracy and enabling clinical applications.

This study benchmarks 26 variations of polygenic risk score models to assess how different assumptions affect disease prediction accuracy using UK Biobank data.

The paper uses systems biology analysis of whole-genome sequences from 298 SB case-control pairs to identify key pathways and genes associated with spina bifida risk.

This study identifies gene-drug interactions that stratify breast cancer risk using polygenic scores and UK Biobank data, finding significant associations with corticosteroid use.

This study uses deep learning to predict chromosomal instability (CIN) in breast cancer from histopathology images, achieving an AUC of 0.822 in a test set.

This paper develops a statistical framework to detect gene-environment interactions by focusing on variance-associated genetic variants, finding higher discovery rates for BMI and disease-related traits.

This study uses single-cell RNA sequencing to analyze SVZ stem cells across aging in mice, identifying distinct cell types and aging-related changes in stem cell behavior.

The paper identifies FBXW7 mutations as a prognostic marker in canine B-cell lymphoma using exome sequencing and clinical data.

This paper evaluates the clinical impact of polygenic risk scores across 23 diseases using UK Biobank data, showing improved prediction accuracy and actionable risk mitigation strategies.

The paper harmonizes clinical interpretations of cancer genomic variants across six knowledgebases, improving overlap and matching with patient cohorts.

The paper identifies three novel complex structural variation patterns in cancer genomes using genome graph analysis across 2,833 tumor WGS.

This study identifies molecular subgroups of BPH and suggests mTOR inhibitors as a potential therapy for specific subtypes.

This paper introduces ECLIPSE, a machine learning model that predicts essential genes and pharmacological targets in cancer by integrating genomic, cell line, and experimental features, showing better differentiation of approved cancer drugs from non-approved ones.

The paper analyzes bulk RNA-seq data to identify factors influencing immune cell infiltration in human tissues, finding associations with age, sex, and genetic variants.

This study uses single-cell RNA sequencing to analyze transcriptional changes in spermatogonia and spermatocytes across developmental stages, revealing age-related gene expression differences and pathway utilization.

This paper develops predictive models for drug synergy using multi-task learning, showing high accuracy in predicting synergistic drug pairs and their clinical relevance.

The paper predicts cancer cell-of-origin from DNA sequences by matching mutational landscapes to chromatin modification patterns in normal tissues.

The paper trains a random forest classifier, ForestMHC, to predict MHC-I peptide presentation using a million human peptides, outperforming existing tools on test sets and mass spectrometry data.

This paper develops an AI system to assess human blastocyst quality using deep learning, achieving 98% accuracy and identifying clinical factors affecting pregnancy rates.

MAESTER predicts tissue-specific drug adverse events by integrating compound structure, targets, phenotypic effects, and genomic data, achieving high accuracy and identifying previously unknown adverse events.

Fine-tuned ImageNet ResNets beat Inception models on tissue micro-array and BreaKHis slides, reaching 99.8% across four tumour tissues and 96.4% across four malignant sub-types.

This study develops a CNN-based pipeline to classify histopathology images across cancer types and biomarkers, achieving high accuracy in distinguishing cancer subtypes and IHC scores.

This paper introduces BANDIT, a Bayesian machine-learning framework for drug target identification, achieving 90% accuracy in predicting targets for 2000 small molecules and validating novel microtubule inhibitors and the target of ONC201.

EthSEQ annotates ethnicity from whole exome sequencing data with high precision, validated on 1,000 Genome Project and TCGA datasets.

This study reports that 17% of advanced cancer patients have clinically relevant germline variants, with higher rates of cancer-associated variants compared to control cohorts.

The paper introduces PMKB, a platform for collaborative editing and sharing of clinical-grade cancer mutation interpretations, and reports its usage in genomic testing.

AGFusion reconstructs and visualizes gene fusion proteins from genomic breakpoints to aid in assessing their functional impact.