Research

A few research lines I’ve been working on: what the question is, what I did, and what came out of it.

PhD project — BCR stereotypy × NOTCH1 in CLL

Question

In CLL, NOTCH1 is one of the most recurrent mutations, and BCR stereotypy defines reproducible biological subsets. Evidence also points to a bidirectional link between BCR signaling and NOTCH1.
I first map the programs that define stereotyped subsets in NOTCH1-wt CLL, then test how NOTCH1 co-occurrence rewires those programs within each subset, and whether the shifts align with the expected activation/aggressiveness differences.

Design (what we compare)

Subsets: #4 (IGHV4-34), #1 (IGHV1-69), #3 (IGHV3-21)
Genotype: NOTCH1-mut vs NOTCH1-wt
Goal: separate subset-intrinsic programs from NOTCH1-dependent effects (and identify where the interaction creates a distinct state)

Why these subsets

Subset #4 (IGHV4-34) is classically linked to a more indolent/anergic phenotype, whereas subset #1 (IGHV1-69) and subset #3 (IGHV3-21) are often associated with more activated/aggressive programs.
The point is to see what’s truly subset-intrinsic and what shifts with genotype and context.

How do I use the data

I use multi-omics to answer biological questions, not just to summarize signals. The aim is to propose a mechanistic explanation (programs/regulators/dependencies), state the predictions it implies, and define the experimental readouts that can confirm or falsify it.

Data layers (in practice)

Bulk RNA-seq — transcriptional programs and pathway-level signatures (BCR-linked signaling; NFAT/AP-1; metabolic/microenvironment modules)
CUT&Tag — TFs and histone marks, including enhancer activity readouts (H3K27ac) and NFAT-centered circuitry
ONT DNA methylation — methylation at promoters/regulatory elements integrated with expression and chromatin state

My contribution

Study design under constraints: multi-batch work (3 batches, dropouts/uneven libraries); rebuilt contrasts when designs became invalid and kept inference interpretable.
QC + robustness: explicit QC gates, batch-aware modeling, and sanity checks so conclusions don’t collapse under reasonable re-analyses.
Mechanistic framing: translate multi-omics output into testable biological claims (not “pathways are enriched”).

Status: ongoing PhD work; links when public.

Richter transformation (first author) — shared signatures to actionable vulnerabilities

Question

RT is rare, aggressive, and heterogeneous. The question behind this study was:
can integrated genomics + transcriptomics identify a reproducible RT “blueprint” across cohorts; and can that blueprint drive a rational drug-repurposing strategy that holds up in RT models?

Study setup (hard data)

20 RT FFPE lymph-node biopsies profiled by targeted DNA sequencing (TSO500)
RNA-seq for 14/20 (FFPE RNA quality constrained feasibility)
Integration with independent published RT cohorts to isolate signals stable beyond a single dataset
Drug-repurposing step built from the multi-omics output: curated 66-drug library (pathway/target-driven) and tested in RT models

My contribution

Co-defined the biological question and study logic with supervisors
Owned the genomics/transcriptomics integration, cross-cohort comparisons, and signature → vulnerability prioritization logic
Led interpretation and manuscript drafting with supervisor feedback

Experimental linkage

Not an in-silico-only story: prioritization was followed by in vitro testing in RT models (RT-PDX-derived models and U-RT1 cell line).

Output

Manuscript (submitted, Leukemia) — Common genomic and transcriptomic signatures in Richter transformation highlight druggable vulnerabilities and guide drug repurposing strategies.
Selected oral presentation — European School of Hematology (ESH), Vienna — Mar 2024

Selected contribution (second author) — BCR–NOTCH1 and metabolic reprogramming in CLL

Outside my PhD line, I contributed heavily to a manuscript on metabolic reprogramming in CLL in the context of BCR–NOTCH1 cooperation, focusing on analysis framework and design/QC decisions.

Output: Functional cooperation between the BCR and NOTCH1 in regulating metabolic reprogramming in chronic lymphocytic leukemia. (submitted/under review; second author)

Selected methods-heavy contribution (outside my main disease area)

Donor-derived cfDNA in pediatric heart transplant monitoring (2024).
I contributed the statistical/ML component, including the random forest analysis used for predictive modeling and feature prioritization.

Reference: Sorbini M, Aidala E, Carradori T, Vallone FE, et al. Donor-derived Cell-free DNA Evaluation in Pediatric Heart Transplant Recipients: A Single-center 12-mo Experience. (2024).

Other ongoing collaborations

Ongoing bulk RNA-seq analyses collaboration with CLL groups at IRCCS San Raffaele Hospital (HSR, Milan), including Scielzo and Ghia labs.