Projects

Enhancing copy number variants (CNVs) detection from sequencing

by Romina D'Aurizio, Elia Ceroni CNV, WES, Gene Panel, Targeted Sequencing, Structural Variant

We develop computational methods for the identification of copy number variants (CNVs), which refer to the presence of an abnormal number of copies of a particular DNA segment. The fraction of the human genome affected by CNVs is substantial (5-10%) and contributes to phenotypic variation, but it also impacts several diseases, including cancer, cardiovascular disease, HIV acquisition and progression, autoimmune diseases, and Alzheimer’s and Parkinson’s diseases in complex and as yet unexplored ways.

A Quantum Enhanced machine learning tool for drug repurposing in rare cancer

by Valeria Repetto Quantum Machine Learning, Rare tumours, Drug repurposing

Classical computational models often fail to capture the complexity of oncological data especially in the case of rare cancer for which there are no standard treatments due to fewer and fragmented data currently available. Machine learning and quantum computing are two technologies with the potential to revolutionise how computation is performed to address previously untenable problems. This project aims to exploit the potential of quantum machine learning algorithms to link new undetected target patterns and results from publicly available cell-line-based pharmacogenomic screens for drug repurposing in rare tumors.

Identification of actionable cancer neoepitopes using just tumor RNAseq

by Danilo Tatoni Transcriptomics, Cancer, Immunotherapy

Cancer vaccination is one of the most appealing strategies to treat solid tumors, and it’s designed as a truly individualized approach. In the current design it still requires the concurrent genomic and transcriptomic profilation of the cancer specimen, making the approach unscalable. Our work is focused on investigating the solely use of the RNAseq profile to identify targetable cancer neoepitopes, decreasing costs and speeding up the development process.

In silico investigation of BRAFV600E regulatory networks in melanoma

by Maurizio Podda

The BRAF protein kinase is widely studied as a cancer driver and therapeutic target. The BRAF V600E mutation is the most frequent BRAF point mutation. The 600 amino acid valine (V) is substituted by glutamic acid (E), which drives diagnoses such as melanoma, papillary thyroid carcinoma, colorectal cancer, and non-small-cell lung. This project aims at understanding whether differences in the ratio between two BRAF(V600E) transcript variants (called ref and X1, PMID: 28454577) are associated with differences in functional signatures, prognosis, and response to treatment.

Model for single exon Copy Number Variant detection in diagnostic

by Giulia Brunelli

We set up a computational pipeline based on EXCAVATOR2 for the detection of CNVs in genetics clinical practice. However at the state of the art, we are not able to identify CNVs shorter than 3 exons. Therefore we aim to exploit the distribution of the ratio between the coverage depth of one sample and the mean coverage of a pool of controls to define the prior of a Bayesian model for single-exon CNV calling.

Modeling 3-Dimensional chromatin interactions rearrangments in Neurodevelopmental Disorders

by Orazio Catona

The project’s aim is to get a better understanding on how Promoter-Enhancer Interactions regulate gene expression and how the disruption of such interactions affects the balance of the regulation, allowing us to discover new genomic hotspots related to diseases. What we want is to define the set of interactions possible for a specific tissue or cell type taking into account their chromatin state and their Topologically Associated Domains (TAD), we want to change this network based on the alterations in a patient to see what can possibly change with Copy Number Variants (CNVs) and rank their pathogenicity and identify how non-coding regions can affect pathogenicity.

SI-MARKERS: A new “SINLAB” AI-based module to support the clinician in the outcome prediction of patients with brain tumors

by Barbara Iadarola Multi-omics Integration, MRI Segmentation, Machine Learning, Personalized Medicine

Many clinical diseases are characterized by extremely complex phenotypes, often determined by multiple variables. In such cases, the totality of the molecular complexity can be poorly explicated by a single element of causality, making necessary the integration of multiple omic groups. The aim of this project is to provide a proof of concepts on the advantage of multi-omics integration throughout an AI-based framework. Imaging, clinical and fluid biomarkers will be integrated, setting-up, fine-tuning and evaluating different machine learning methods for the prediction of the clinical outcome of the patient.