Our research 

We work in computational oncology, an interdisciplinary field that combines informatics, mathematics, physics, and statistics with oncology. We develop new tools to understand how tumours evolve and respond to treatment, with the aim of predicting disease dynamics and improving precision medicine

Members of our lab have heterogeneous training, ranging from STEMS to Life Sciences. To develop our tools we integrate bioinformatics, mathematics and statistical models based on population genetics and machine learning. The final output is often software that we use to approach real-world biological problems such as those discussed below.

Over the years, we have established a large network of experimental and computational collaborators across universities, research centres and hospitals. Among them, we have some of the world-leading research groups in our field. 

Three research problems that we like

Pre-existing versus de novo treatment resistance

The emergence of a population of cancer cells that survive treatment is disruptive for a patient. Since all cells in the human body are evolutionarily related (and so do the resistant ones), it makes sense to ask whether the first resistant cell was already present (pre-existing), or not (de novo) before treatment. 

This very simple question is extremely challenging, but is important to understand disease resistance dynamics. In collaboration with Luca Vago and Giovanni Tonon (San Raffaele Hospital, Milan), we are developing new tools to measure treatment resistance from longitudinal sequencing data of acute myeloid leukaemia, one of the most lethal haematological cancers.

Genetic, epi-genetic evolution and plasticity

Cancer cells evolve by acquiring genetic and epigenetic mutations that can enhance cellular plasticity, i.e. the ability to change behaviour in response to external or internal signals. Plasticity allows cancer cells to adapt to different environments, evade the immune system, and resist therapies.

In collaboration with Valter Gattei (National Cancer Centre, Aviano), we are developing new tools to measure plasticity from longitudinal sequencing data of chronic lymphocytic leukaemia, one of the most well-studied haematological cancers.

Evolutionary trajectories across time and space

New technologies allow measuring tumours in time and space at higher resolution. Whether we look at the DNA, RNA or chromatin, we get a partial perspectives on the evolutionary process hidden in the individual and, across multiple patients, we see the process at the cohort level. 

In collaboration with Andrea Sottoriva (Human Technopole, Milan), Trevor Graham (Institute of Cancer Research, London), Alona Sosinsky (Genomics England, UK), and the Laboratory of Data Engineering (Area Science Park, Trieste) we develop new tools to measure spatio-termporal evolution from new sequencing technologies and large-scale cohorts.

Our hands-on experience

Machine Learning

We regularly use probabilistic programming languages such as Pyro and STAN to implement machine learning models for clustering, regression, features selection and dimensionality reduction. 

Next-generation sequencing 

We have experience with the most popular sequencing technologies (Illumina, Nanopore, 10x, etc.). We have used them to measure DNA from bulk whole-genome/exome assays, or to measure RNA and ATAC from single cells, even in multi-omics and spatial assays.

Experimental model systems

We have experience with canonical tumour-normal and tumour-only designs, as well as with patient-derived organoids. We are often involved from designing the experiments to analysing the data and, in our projects, we generate our own sequencing data!

Where are we

We are based in Trieste, a city of ~200.000 inhabitant in the north-east corner of Italy, next to Slovenia and Croatia. 

Trieste is a seaside city enclosed by mountains, with beautiful landscapes and many outdoor activities. It has high quality of life, low cost of living and a mediterranean vibe.