Biology subjects

Giacopuzzi, E.

Publications and source records attributed to Giacopuzzi, E..

8 recordsLinked to original sources

bioRxiv2026

Recurrent leukaemia stem cell states define the antigenic landscape of paediatric Acute Myeloid Leukaemia

Paediatric acute myeloid leukaemia (pAML) remains a leading cause of childhood cancer mortality, with disease heterogeneity representing a major obstacle to effective immunotherapy. To enable rational immunotherapy design accounting for such heterogeneity, we combined single-cell transcriptomics and surface proteomics (CITE-seq). Using an integrated framework for malignant cell identification and unbiased gene program identification, we resolved three conserved leukaemia stem cell (LSC) subtypes, primitive (P-LSC), monocytic (M-LSC), and erythroid-like (E-LSC), that frequently coexist within individual patients in varying proportions and represent distinct stem-cell states with characteristic immunophenotypes and transcriptional programmes. Quantitative surface antigen profiling revealed that no single target achieved full LSC coverage. CLEC12A, an antigen currently being evaluated clinically, was expressed across P-LSC and M-LSC but was largely absent from E-LSC. In contrast, CD84 provided complementary coverage in the P and M groups and served as the primary target in the E-LSC population. The combination of CLEC12A and CD84 provided broad coverage across all three LSC subtypes, while sparing healthy haematopoietic stem cells. Validation across independent single-cell and bulk cohorts confirmed the robustness of this combination. These findings establish a paediatric-specific framework for antigen prioritisation in AML, identify CLEC12A/CD84 as a robust dual-targeting strategy across heterogeneous LSC states, and provide a blueprint for the rational design of next-generation CAR T-cell combinations.

cancer biology↗

bioRxiv2025

Competing gene regulatory networks drive naive and memory B cell differentiation

Elucidating the gene regulatory networks (GRNs) that govern human B cell differentiation is essential for understanding immune responses to infection, vaccination and autoantigens. Here, we show that individual naive B cells can give rise to both plasma cells and germinal centre (GC) B cells. In contrast, memory B cells display a progressive increase in IRF4 activity over time, leading to PRDM1 induction and exclusive differentiation into plasma cells. Using CRISPR-based perturbations, we demonstrate that IRF4 is indispensable for both GC and plasma cell development. Notably, IRF4 promotes GC fate independently of PRDM1, as PRDM1 disruption did not impair GC differentiation. We also find that while the abundance of antibody mRNAs is clonally correlated, class switch recombination (CSR) occurs stochastically and is clonally independent. Together, these findings reveal distinct regulatory dynamics during naive and memory B cell activation and offer new insights into the GRNs underlying human B cell fate decisions.

immunology↗

bioRxiv2023

Self-supervised learning for characterising histomorphological diversity and spatial RNA expression prediction across 23 human tissue types

As vast histological archives are digitised, there is a pressing need to be able to associate specific tissue substructures and incident pathology to disease outcomes without arduous annotation. Such automation provides an opportunity to learn fundamental biology about how tissue structure and function varies in a population. Recently, self-supervised learning has proven competitive to supervised machine learning approaches in classification, segmentation and representation learning. Here, we leverage self-supervised learning to generate histology feature representations using 1.7M images across 23 healthy tissues in 838 donors from GTEx. Using these representations, we demonstrate we can automatically segment tissues into their constituent tissue substructures and pathology proportions, and surpass the performance of conventionally used pre-trained models. We observe striking population variability in canonical tissue substructures, highlight examples of missing pathological diagnoses, incorrect assignment of target tissue and cross-tissue contamination. We demonstrate that this variability in tissue composition leads to a likely overestimation of eQTL tissue sharing and drives dramatic differential gene expression changes. We use derived tissue substructures to detect 284 tissue substructures and pathology specific eQTLs. As our derived histology representations are rich morphological descriptors of the underlying tissue, we introduce a multiple instance learning model that can predict and spatially localise individual RNA expression levels directly from histology to specific substructures and pathological features. We validate our RNA spatial predictions with matched ground truth immunohistochemistry (IHC) for several well characterised marker genes, recapitulating their known spatial specificity. Finally, we derive a gene expression spatial enrichment metric, allowing us to detect genes specifically expressed within sites of pathology (e.g. arterial calcification). Together, these results demonstrate the power of self-supervised machine learning when applied to vast histological datasets to allow researchers to pose and answer questions about tissue pathology, its spatial organisation and the interplay between morphological tissue variability and gene expression.

bioinformatics↗

bioRxiv2023

CHARR efficiently estimates contamination from DNA sequencing data

DNA sample contamination is a major issue in clinical and research applications of whole genome and exome sequencing. Even modest levels of contamination can substantially affect the overall quality of variant calls and lead to widespread genotyping errors. Currently, popular tools for estimating the contamination level use short-read data (BAM/CRAM files), which are expensive to store and manipulate and often not retained or shared widely. We propose a new metric to estimate DNA sample contamination from variant-level whole genome and exome sequence data, CHARR, Contamination from Homozygous Alternate Reference Reads, which leverages the infiltration of reference reads within homozygous alternate variant calls. CHARR uses a small proportion of variant-level genotype information and thus can be computed from single-sample gVCFs or callsets in VCF or BCF formats, as well as efficiently stored variant calls in Hail VDS format. Our results demonstrate that CHARR accurately recapitulates results from existing tools with substantially reduced costs, improving the accuracy and efficiency of downstream analyses of ultra-large whole genome and exome sequencing datasets.

bioinformatics↗

bioRxiv2020

GREEN-DB: a framework for the annotation and prioritization of non-coding regulatory variants in whole-genome sequencing

Non-coding variants have emerged as important contributors to the pathogenesis of human diseases, not only as common susceptibility alleles but also as rare high-impact variants. Despite recent advances in the study of regulatory elements and the availability of specialized data collections, the systematic annotation of non-coding variants from genome sequencing remains challenging. Here, we propose a new framework for the prioritization of non-coding regulatory variants that integrates information about regulatory regions with prediction scores and HPO-based prioritization. Firstly, we created a comprehensive collection of annotations for regulatory regions including a curated database of 2.4 million regulatory elements (GREEN-DB) annotated with controlled gene(s), tissue(s) and associated phenotype(s) where available. Secondly, we calculated a variation constraint metric and showed that constrained regulatory regions associate with disease-associated genes and essential genes from mouse knock-out screens. Thirdly, we compared 19 non-coding impact prediction scores providing suggestions for variant prioritization. Finally, we developed a VCF annotation tool (GREEN-VARAN) that can integrate all these elements to annotate variants for their potential regulatory impact. In our evaluation, we show that GREEN-DB can capture previously published disease-associated non-coding variants as well as identify additional candidate disease genes in WGS trio analyses.

bioinformatics↗

bioRxiv2020

Investigating an in-silico approach for prioritizing antidepressant drug prescription based on drug-induced expression profiles and predicted gene expression

In clinical practice, antidepressant prescription is a trial and error approach, which is time consuming and discomforting for patients. This study investigated an in-silico approach for ranking antidepressants based on their hypothetical likelihood of efficacy. We determined the transcriptomic profile of citalopram remitters by performing a transcriptomic-wide association study on STAR*D data (N =1163). The transcriptional profile of remitters was compared with 21 antidepressant-induced gene expression profiles in five human cell lines available in the connectivity map database. Spearman correlation, Pearson correlation, and the Kolmogorov Smirnov test were used to determine the similarity between antidepressant-induced profiles and remitter profiles, subsequently calculating the average rank of antidepressants across the three methods and a p-value for each rank by using a permutation procedure. The drugs with the top ranks were those having high positive correlation with the expression profiles of remitters and they may have higher chances of efficacy in the tested patients. In MCF7 (breast cancer cell line), escitalopram had the highest average rank, with an average rank higher than expected by chance (p=0.0014). In A375 (human melanoma) and PC3 (prostate cancer) cell lines, escitalopram and citalopram emerged as the second highest ranked antidepressants, respectively (p=0.0310 and 0.0276, respectively). In HA1E (kidney) and HT29 (colon cancer) cell types, citalopram and escitalopram did not fall among top antidepressants. The correlation between citalopram remitters and (es)citalopram-induced expression profiles in three cell lines suggests that our approach may be useful and with future improvements it can be applicable at the individual level to tailor treatment prescription.

bioinformatics↗

bioRxiv2018

Genome-wide analysis of RNA editing levels in human blood identified interactions with mRNA processing genes and suggested correlations with biological and drug-related variables

BackgroundA-to-I RNA editing is a co-/post-transcriptional modification catalyzed by ADAR enzymes, that deaminates Adenosines (A) into Inosines (I). Most of known editing events are located within inverted ALU repeats, but they also occur in coding sequences and may alter the function of encoded proteins. RNA editing contributes to generate transcriptomic diversity and it is found altered in cancer, autoimmune and neurological disorders. Emerging evidences indicate that editing process could be influenced by genetic variations, biological and environmental variables.\n\nResultsWe analyzed RNA editing levels in human blood using RNA-seq data from 459 healthy individuals and identified 2,079 sites consistently edited in this tissue. As expected, analysis of gene expression revealed that ADAR is the major contributor to editing on these sites, explaining [~]13% of observed variability. After removing ADAR effect, we found significant associations for 1,122 genes, mainly involved in RNA processing. These genes were significantly enriched in genes encoding proteins interacting with ADARs, including 276 potential ADARs interactors and 9 ADARs direct partners. In addition, our analysis revealed several factors potentially influencing RNA editing in blood, including cell composition, age, Body Mass Index, smoke and alcohol consumption. Finally, we identified genetic loci associated with editing levels, including known ADAR eQTLs and a small region on chromosome 7, containing LOC730338, a lincRNA gene that appears to modulate ADARs mRNA expression.\n\nConclusionsOur data provides a detailed picture of the most relevant RNA editing events and their variability in human blood, giving interesting insights on potential mechanisms behind this post-transcriptional modification and its regulation in this tissue.

genomics↗

bioRxiv2017

GARFIELD-NGS: Genomic vARiants FIltering by dEep Learning moDels in NGS

SummaryExome sequencing approach is extensively used in research and diagnostic laboratories to discover pathological variants and study genetic architecture of human diseases. However, a significant proportion of identified genetic variants are actually false positive calls, and this pose serious challenges for variants interpretation. Here, we propose a new tool named GARFIELD-NGS (Genomic vARiants FIltering by dEep Learning moDels in NGS), which rely on deep learning models to dissect false and true variants in exome sequencing experiments performed with Illumina or ION platforms. GARFIELD-NGS showed strong performances for both SNP and INDEL variants (AUC 0.71 - 0.98) and outperformed established hard filters. The method is robust also at low coverage down to 30X and can be applied on data generated with the recent Illumina two-colour chemistry. GARFIELD-NGS processes standard VCF file and produces a regular VCF output. Thus, it can be easily integrated in existing analysis pipeline, allowing application of different thresholds based on desired level of sensitivity and specificity.\n\nAvailabilityGARFIELD-NGS available at https://github.com/gedoardo83/GARFIELD-NGS\n\nContactedoardo.giacopuzzi@unibs.it

bioinformatics↗