Giulio Caravagna

This paper introduces a Bayesian framework for differential gene expression analysis that scales to over 10 million cells and integrates data from multiple patients.

This paper introduces SCOUT, a synthetic whole-genome sequencing resource for benchmarking cancer genomic analysis, revealing performance issues in complex tumours and highlighting the importance of evolutionary ground truth.

DeConveil extends differential gene expression analysis to account for genome aneuploidy by integrating CNV data, improving gene classification and biomarker discovery in cancer.

BASCULE uses a Bayesian framework to standardize and cluster mutational signatures across cancer types, identifying prognostic subtypes in over 8,000 samples.

The paper introduces ctDNAmer, a reference-free tool for detecting ctDNA from WGS data by analyzing k-mers, showing improved tumor fraction estimation and detection compared to traditional methods.

The paper shows how epigenetic heritability of cell plasticity contributes to cancer drug resistance through a one-to-many genotype-to-phenotype mapping.

This paper introduces MOBSTERm, a Bayesian framework for subclonal deconvolution that accounts for spatio-temporal sampling biases in cancer sequencing data, improving the analysis of tumour evolution.

This paper introduces TickTack, a Bayesian model to time and cluster co-occurring genome amplifications in cancer, revealing insights into tumor evolution and genomic instability.

This paper introduces a Bayesian framework to predict tumour evolution and drug resistance using longitudinal cancer data across three cancer types.

This paper introduces a high-throughput microfluidic platform for organoid culture and multi-omics profiling, demonstrating its utility in clinical and drug discovery settings.

This paper introduces MIDAA, a framework for multi-omic data integration using deep learning and biological principles, showing it identifies interpretable and biologically relevant patterns better than existing methods.

The paper develops a tool to assess tumour contamination in normal samples using whole-genome sequencing data, finding contamination in various cancer types and highlighting risks to variant calling.

CONGAS+ is a Bayesian method that maps single-cell RNA and ATAC data to copy number clones, revealing subclonal architectures and genotype-phenotype relationships in cancer.

This study identifies cancer type-specific evolutionary signatures using genomic data from over 35,000 patients, revealing patterns of driver mutation acquisition linked to prognosis.

CONGAS clusters single-cell RNA sequencing data using copy number alterations to infer subclonal composition and RNA phenotypes in cancer data.

PMCE infers probabilistic graphical models of cancer evolution from mutational profiles, showing a significant correlation between predicted evolutionary paths and patient survival.

The paper analyzes AID-related mutations across cancers and finds they are associated with better outcomes in ICI-treated patients and contribute to immunogenic neoepitopes.

This paper introduces CNAqc, a toolset for validating clonal and subclonal copy number alterations in bulk tumor sequencing by addressing tumor purity, ploidy, and heterogeneity using simulations and single-cell data.

This paper introduces a model-based subclonal reconstruction method that integrates evolutionary theory with machine learning to improve accuracy in tumor subclonal analysis.

The paper quantifies tumor immunogenicity using dN/dS ratios in the immunopeptidome and shows it predicts immunotherapy response by distinguishing adaptation from escape mechanisms.

This paper shows that spatial tumour structure affects mutation patterns in sequencing data and introduces a framework to account for these effects in cancer genomic analysis.

The paper introduces two Cytoscape-based tools, cyTRON and cyTRON/JS, for inferring cancer evolution models from genomic and proteomics data.

TRaIT infers mutational graphs from single-cell and multi-region sequencing data to model tumour evolution, improving accuracy and robustness.

This paper introduces a machine learning method to detect repeated cancer evolution patterns across patients using multi-region sequencing data, identifying prognostic patient subgroups.

This paper introduces Suppes-Bayes Causal Networks to model the accumulation of DNA changes in diseases like cancer and HIV, showing their ability to infer causal relationships from genetic data.

The paper introduces PiCnIc, a machine learning pipeline to infer cancer progression models from NGS data, demonstrating its application to colorectal cancer and suggesting novel hypotheses.

This paper introduces a statistical machine learning method to reconcile model predictions across different abstraction levels in biological systems.

TRONCO infers cancer progression models from genomic data, providing population-level and individual-level insights into clonal evolution and mutation accumulation.

TRONCO is an R package for cancer progression modeling using cross-sectional genomic data, offering improved algorithms and tools for analysis and visualization.