Mattia Prosperi

Vir2vec is a genome-wide viral embedding model that outperforms existing models in viral genome understanding tasks and provides explainability for host adaptation mutations.

This study identifies socioeconomic and clinical factors influencing BRCA genetic testing uptake among cancer patients using electronic health records and develops a predictive model to identify those at risk for underuse.

AURA is a scalable machine learning framework for predicting antibiotic resistance in Mycobacterium tuberculosis, identifying novel resistance-associated loci.

SARITA is a large language model that generates synthetic SARS-CoV-2 Spike S1 sequences, accurately predicting mutations in future variants like Delta and Omicron.

This paper develops a framework combining paired-sample dynamics, pathway-based analytics, and MLOps to improve transcriptomic classification in small cohorts, achieving high precision and recall in case studies.

This paper introduces DeepAutoCoV, a deep learning system that predicts future dominant SARS-CoV-2 lineages by detecting anomalies in genomic data, achieving better performance than baseline methods with early detection of emerging variants.

This paper develops a machine learning classifier to predict antibiotic resistance in Mycobacterium tuberculosis using genomic data and achieves high accuracy for first-line antibiotics.

DeepDynaForecast uses phylogenetic trees and graph deep learning to predict epidemic transmission dynamics, showing accuracy in simulated and real HIV data.

OCTOPUS is a metagenomics classifier for portable sequencing data that improves accuracy and reduces false positives on mobile devices.

KARGVA identifies antibiotic resistance gene variants from metagenomics data with high accuracy and efficiency, outperforming existing tools in detecting ARGVs.

The paper evaluates AMR prediction methods using HTS data from clinical isolates, finding high variance in performance and highlighting issues with sampling bias and uncharacterized AMR genes.

TARDiS optimizes viral genome subsampling by balancing genetic diversity and temporal distribution to improve phylogenetic analysis.

The paper introduces a Gaussian mixture model clustering method to find overlapping Rmaps without quantization, improving precision and efficiency in optical mapping analysis.

DYNAMITE identifies dynamic transmission clusters in viral phylogenies by analyzing branches rather than clades, outperforming existing tools in simulated outbreaks.

This study analyzes how global and regional public health interventions affected SARS-CoV-2 transmission clusters using genetic sequence data and phylogenetic analysis.