Biology subjects

Milite, S.

Publications and source records attributed to Milite, S..

11 recordsLinked to original sources

bioRxiv2026

Tumour evolution as ground truth for cancer whole-genome sequencing

Cancer genomes are shaped by evolutionary processes that couple mutagenesis, clonal selection, chromosomal instability, spatial growth and treatment response into structured genomic patterns, yet current benchmarking strategies largely ignore this evolutionary dependency. Here, we present SCOUT, a large-scale synthetic whole-genome sequencing resource of over 200 samples, designed for systematic benchmarking of tumour genomic analysis and evolutionary inference under controlled evolutionary ground truth. Unlike conventional task-specific simulations, SCOUT models tumour evolution as a latent generative process that simultaneously shapes mutations, copy-number alterations, variant allele frequencies, mutational signatures and clonal architectures. SCOUT recapitulates key features of solid and haematological malignancies, including driver mutations, chromosomal instability, intratumour heterogeneity, spatial sampling and treatment-associated evolutionary dynamics in tumour and matched-normal longitudinal and multi-region sequencing designs. Using SCOUT, we benchmarked widely used methods for somatic variant detection, copy-number analysis, mutational signature inference and tumour evolutionary reconstruction. Across analytical tasks, performance deteriorated in low-purity, highly subclonal and structurally complex tumours, while spatial sampling bias and hypermutation generated spurious evolutionary signals that confounded tumour interpretation across multiple inference layers. Evolutionary simulations further distinguished lineage-restricted genetic bottlenecks from multi-lineage resistance dynamics associated with tumour plasticity. Tumour purity consistently exerted a stronger effect on inference accuracy than sequencing depth. Together, our results establish evolutionary ground truth as a prerequisite for reproducible benchmarking and biologically interpretable analysis of cancer whole-genome sequencing data.

bioinformatics↗

bioRxiv2025

Scalable, fast and accurate differential gene expression testing from millions of cells of multiple patients

Since the development of DNA microarrays and later RNA bulk sequencing, testing with statistically independent samples has been the standard method for detecting genes with different transcription patterns. Single-cell assays challenge these assumptions because individual cells are statistically dependent, and all proposed methodologies present mathematical limitations or computational bottlenecks that prevent a seamless integration of data from many cells and patients simultaneously. In this work, we solve this crucial limitation by introducing a Bayesian framework that retrieves the independence structure at the level of individual patients, separating differences across individuals from actual transcriptional differences. Leveraging multi-GPU and variational inference, our approach excels across different experimental designs and scales to analyse over 10 million cells. This framework enables single-cell differential expression analysis that can finally integrate datasets from large clinical cohorts, atlas projects, or drug-response screens with thousands of samples and millions of cells.

bioinformatics↗

bioRxiv2025

Model-based tumour subclonal deconvolution accounting for spatio-temporal sampling biases

Bulk DNA sequencing has entered the clinic, and understanding tumour evolution in space and time has become critical for advancing precision oncology. A recent work by us has brought a population genetics perspective into the classical tumour subclonal deconvolution problem, showing how multiple spatiotemporal sampling biases arise when we collect more than one sample of the same tumour. Sampling biases complicate the mapping of the mutations into the evolutionary process, a complex issue undermined by existing multi-sample analysis methods. This work presents MOBSTERm, a novel mixture-within-mixture Bayesian framework that extends our earlier approach to a multi-dimensional formulation. Our model incorporates distinct mathematical distributions that capture sampling bias patterns in tumour data, allowing the deconvolution to resist the effect of some confounders. This works presents the simulation of one source of sampling bias analysed with this new approach, together with a large scale test based on simulations. Moreover, we apply our model to understand subclonal deconvolution from a multi-region whole-genome colorectal cancer sample and from longitudinal whole-genome glioblastoma samples of 15 patients collected before and after treatment. This new deconvolution approach offers a better account of the effect of spatio-temporal tumour biases, allowing us to better elucidate complex clonal dynamics from multi-sample cancer sequencing data. Code availabilityMOBSTERm is available at the GitHub page https://github.com/caravagnalab/MOBSTERm. The code to reproduce our analyses is available at Zenodo https://doi.org/10.5281/zenodo.14867256.

cancer biology↗

bioRxiv2024

A Bayesian framework to infer and cluster mutational signatures leveraging prior biological knowledge.

Mutational signatures provide key insights into cancer mutational processes, but the availability of signature catalogues generated by different groups using distinct methodologies underscores a need for standardisation. We introduce a Bayesian framework that offers a systematic approach to expanding existing signature catalogues for any type of mutational signature, while grouping patients based on shared signature patterns. We demonstrate that this approach can identify both known and novel molecular subtypes across nearly 8,000 samples spanning six cancer types, and show that stratifications derived from signature yield prognostic groups, further enhancing the translational potential of mutational signatures.

cancer biology↗

bioRxiv2024

Interpretable Multi-Omics Data Integration with Deep Archetypal Analysis

High-throughput multi-omic molecular profiling allows probing biological systems at unprecedented resolution. However, the integration and interpretation of high-dimensional, sparse, and noisy multimodal datasets remains challenging. Deriving new biology using current methods is particularly difficult because they are not based on biological principles, but instead focus exclusively on a dimensionality reduction task. Here we introduce MIDAA (Multiomic Integration with Deep Archetypal Analysis), a framework that combines archetypal analysis, an approach grounded in biological principles, with deep learning. Using the concept of archetypes that are based on evolutionary trade-offs and Pareto optimality - MIDAA finds extreme data points that define the geometry of the latent space, preserving the complexity of biological interactions while retaining an interpretable output. We demonstrate that indeed these extreme points represent cellular programmes reflecting the underlying biology. We show on real and simulated multi-omics data how MIDAA outperforms state-of-the-art methods in identifying parsimonious, interpretable, and biologically relevant patterns.

bioinformatics↗

bioRxiv2023

NeuroVelo: interpretable learning of cellular dynamics from single-cell transcriptomic data

Reconstructing temporal cellular dynamics from static single-cell transcriptomics remains a major challenge. Methods based on RNA velocity are useful, but interpreting their results to learn new biology remains difficult, and their predictive power is limited. Here we propose NeuroVelo, a method that couples learning of an optimal linear projection with non-linear Neural Ordinary Differential Equations. Unlike current methods, it uses dynamical systems theory to model biological processes over time, hence NeuroVelo can identify gene interactions that drive the observed temporal dynamics of gene expression. We benchmark NeuroVelo against several state-of-the-art methods using single-cell datasets, demonstrating that NeuroVelo simultaneously reconstructs correct cell-type transitions and identifies gene regulatory networks that drive cell fate directly from the data.

bioinformatics↗

bioRxiv2023

Epigenetic heritability of cell plasticity drives cancer drug resistance through one-to-many genotype to phenotype mapping

Cancer drug resistance is multi-factorial, driven by heritable (epi)genetic changes but also phenotypic plasticity. Here we dissect it by perturbing colorectal cancer patient-derived organoids longitudinally with drugs in sequence. Combining longitudinal tracking, single cell omics, evolutionary modelling, and machine leaning, we found that different targeted drugs select for distinct subclones, supporting rationally designed drug sequences. The cellular memory was encoded as a heritable epigenetic configuration, from which multiple transcriptional programmes could run, supporting a one-to-many (epi)genotype-to- phenotype map that explains how clonal expansions and plasticity manifest together. This may ensure drug resistance subclones can exhibit distinct phenotypes in changing environments while still preserving the cellular memory encoding for their selective advantage. Chemotherapies resistance was instead entirely driven by plasticity. Inducing further chromosomal instability before drug application changed clonal evolution but not convergent transcriptional programmes. Collectively, our data show how genetic and epigenetic alterations are selected to "permissive epigenome" enabling phenotypic plasticity.

cancer biology↗

bioRxiv2023

A Bayesian method to infer copy number clones from single-cell RNA and ATAC sequencing

Single-cell RNA and ATAC sequencing technologies allow one to probe expression and chromatin accessibility states as a proxy for cellular phenotypes at the resolution of individual cells. A key challenge of cancer research is to consistently map such states on genetic clones, within an evolutionary framework. To this end we introduce CONGAS+, a Bayesian model to map single-cell RNA and ATAC profiles generated from independent or multimodal assays on the latent space of copy numbers clones. CONGAS+ can detect tumour subclones associated with aneuploidy by clustering cells with the same ploidy profile. The framework is implemented in a probabilistic language that can scale to analyse thousands of cells thanks to GPU deployment. Our tool exhibits robust performance on simulations and real data, highlighting the advantage of detecting aneuploidy from two distinct molecules as opposed to other single-molecule models, and also leveraging real multi-omic data. In the application to prostate cancer, lymphoma and basal cell carcinoma, CONGAS+ did retrieve complex subclonal architectures while providing a coherent mapping among ATAC and RNA, facilitating the study of genotype-phenotype mapping, and their relation to tumour aneuploidy. Author summaryAneuploidy is a condition caused by copy number alterations (CNAs), which brings cells to acquire or lose chromosomes. In the context of cancer progression and treatment response, aneuploidy is a key factor driving cancer clonal dynamics, and measuring CNAs from modern sequencing assays is therefore important. In this framing, we approach this problem from new single-cell assays that measure both chromatin accessibility and RNA transcripts. We model the relation between single-cell data and CNAs and, thanks to a sophisticated Bayesian model, we are capable of determining tumour clones from clusters of cells with the same copy numbers. Our model works when input cells are sequenced independently for both assays, or even when modern multi-omics protocols are used. By linking aneuploidy to gene expression and chromatin conformation, our new approach provides a novel way to map complex genotypes with phenotype-level information, one of the missing factors to understand the molecular basis of cancer heterogeneity.

bioinformatics↗

bioRxiv2022

Clinical application of tumour in normal contamination assessment from whole genome sequencing

The unexpected contamination of normal samples with tumour cells reduces variant detection sensitivity, compromising downstream analyses in canonical tumour-normal analyses. Leveraging whole-genome sequencing data available at Genomics England, we develop a tool for normal sample contamination assessment, which we validate in silico and against minimal residual disease testing. From a systematic review of 771 patients with haematological malignancies and sarcomas, we find contamination across a range of cancer clinical indications and DNA sources, with highest prevalence in saliva samples from acute myeloid leukaemia patients, and sorted CD3+ T-cells from myeloproliferative neoplasms. Further exploration reveals 108 hotspot mutations in genes associated with haematological cancers at risk of being subtracted by standard variant calling pipelines. Our work highlights the importance of contamination assessment for accurate somatic variants detection in research and clinical settings, especially with large-scale sequencing projects being utilised to deliver accurate data from which to make clinical decisions for patient care.

bioinformatics↗

bioRxiv2022

Mutation Rate Evolution Drives Immune Escape In Mismatch Repair-Deficient Cancer

Mutation rate optimisation drives evolution and immune evasion of bacteria and lentiviral strains, including HIV. Whether evolving cancer lineages similarly adapt mutation rates to increase tumour cell fitness is unknown. Here, by mapping the clonal topography of mismatch repair-deficient (MMRd) colorectal cancer, we show that genomic MMRd mutability co-evolves with neoantigen selection to drive intratumour diversification and immune escape. Mechanistically, we find that microsatellite instability modulates subclonal DNA repair by toggling two hypermutable mononucleotide homopolymer runs in the mismatch repair genes MSH6 and MSH3 (C8 and A8, respectively) through stochastic frameshift switching. Spontaneous mutation and reversion at these evolvability switches modulates subclonal mutation rate, mutation bias, and clonal HLA diversity during MMRd cancer evolution. Combined experimental and simulation studies demonstrate that subclonal immune selection favours incremental MMR mutations. MMRd cancers thus fuel intratumour heterogeneity by adapting subclonal mutation rate and mutation bias to immune selection, revealing a conserved co-evolutionary arms race between neoantigen selection and adaptive genomic mutability. Our work reveals layers of mutational complexity and microsatellite biology in MMRd cancer evolution previously hidden in bulk analyses.

cancer biology↗

bioRxiv2021

Genotyping Copy Number Alterations from single-cell RNA sequencing

MotivationCancers are composed by several heterogeneous subpopulations, each one harbouring different genetic and epigenetic somatic alterations that contribute to disease onset and therapy response. In recent years, copy number alterations leading to tumour aneuploidy have been identified as potential key drivers of such populations, but the definition of the precise makeup of cancer subclones from sequencing assays remains challenging. In the end, little is known about the mapping between complex copy number alterations and their effect on cancer phenotypes. ResultsWe introduce CONGAS, a Bayesian probabilistic method to phase bulk DNA and single-cell RNA measurements from independent assays. CONGAS jointly identifies clusters of single cells with subclonal copy number alterations, and differences in RNA expression. The model builds statistical priors leveraging bulk DNA sequencing data, does not require a normal reference and scales fast thanks to a GPU backend and variational inference. We test CONGAS on both simulated and real data, and find that it can determine the tumour subclonal composition at the single-cell level together with clone-specific RNA phenotypes in tumour data generated from both 10x and Smart-Seq assays. AvailabilityCONGAS is available as 2 packages: CONGAS (https://github.com/caravagnalab/congas), which implements the model in Python, and RCONGAS (https://caravagnalab.github.io/rcongas/), which provides R functions to process inputs, outputs, and run CONGAS fits. The analysis of real data and scripts to generate figures of this paper are available via RCONGAS; code associated to simulations is available at https://github.com/caravagnalab/rcongas_test. Contactgcaravagna@units.it Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗