Leading the Quantitative Biology Group — computational approaches to decode how cancer cells respond, adapt, and resist treatment.
Our group develops computational approaches to analyse high-throughput drug and genetic screens alongside multi-omics data, with the goal of uncovering the regulatory mechanisms that drive how cancer cells respond to treatment.
Working closely with experimental and clinical collaborators, we build integrative methods to tackle one of oncology's most pressing challenges — drug resistance — while training a new generation of engineers at the intersection of computer science, cell biology, and biomedicine.
Designing and analysing genome-wide CRISPR screens — spanning loss-of-function knockouts, base editing and saturation mutagenesis — to map cancer vulnerabilities and resolve the functional consequences of genetic variation across hundreds of cell models.
Classical and deep generative methods — including variational autoencoders — for integrating genomics, transcriptomics, proteomics, and drug response into unified representations that reveal biologically meaningful structure.
Building and expanding the Cancer Dependency Map: prioritising therapeutic targets, benchmarking biomarkers of response, and extending these maps with synthetic data to cover under-represented cancer contexts.
Partnering with hospitals and registries to bring AI to real-world oncology — extracting structured knowledge from unstructured electronic health records and developing diagnostic and prognostic tools for precision medicine.
A forward-looking review of how synthetic lethality, two decades after its conception, is finally maturing into a practical framework for precision oncology — from screen design to clinical translation.
A comprehensive survey of methods for integrating complex multi-omics datasets, with a focus on variational autoencoders and other deep generative models for imputation, joint embedding, and batch correction.
MOSA — a multi-view variational autoencoder that integrates and synthetically augments the Cancer Dependency Map, increasing the number of multi-omic profiles by 32.7 % across 1,523 cancer cell lines.
Assistant Professor at Instituto Superior Técnico and Integrated Researcher at INESC-ID. PhD in Systems Biology (Cambridge / EMBL-EBI, 2017). Former postdoctoral fellow at the Wellcome Sanger Institute and Fulbright Scholar at the Broad Institute of MIT and Harvard.
Genetic interactions, synthetic lethality and complexity in cancer vulnerability mapping.
A new dual guide CRISPR system for genetic interaction library screening.
Mechanistic insights into the differentiation of CF epithelia — Italian Cystic Fibrosis Research Foundation.
Synthetic lethality in cancer drug discovery — challenges and opportunities.
From classical statistical methods to modern deep generative models.
A pan-cancer vulnerability identified through combinatorial CRISPR screens.
Mosaic integration and synthetic generation of multi-omic data for precision cancer medicines.
Multi-domain open MALDI spectra bank for identification of microorganisms.
AI-driven integration of sarcoma clinical data for the Portuguese National Oncology Registry.
Joining the ELLIS Lisbon Unit within the European Laboratory for Learning and Intelligent Systems.
First synthetic Dependency Map generated with unsupervised deep learning.
Visiting the Broad Institute of MIT and Harvard — deep learning for combinatorial saturation mutagenesis screens.
Second-generation map of cancer dependencies across 930 cell lines.
Mapping mutations affecting interferon-gamma signalling.
Despite remarkable progress in cancer precision medicine, drug resistance and low clinical success rates persist. SYNTHESIS builds synthetic and mosaic-integration frameworks to expand the scope of multi-omic discovery.
An AI-based system to process unstructured electronic health records efficiently, automating the extraction of structured clinical information for the Portuguese National Oncology Registry.
Understanding the molecular mechanisms that govern cystic fibrosis epithelial differentiation, aiming to restore the regenerative balance in CF airways.
Building an open, multi-domain repository of MALDI mass spectrometry data to advance rapid and accurate microbial identification at scale.
