Biology subjects

Cerulo, L.

Publications and source records attributed to Cerulo, L..

6 recordsLinked to original sources

bioRxiv2023

Robust Feature Selection strategy detects a panel of microRNAs as putative diagnostic biomarkers in Breast Cancer

MicroRNAs represent a comprehensive class of short, single-stranded, non-coding RNA transcripts able to interfere with the translation of their targets or degrade them. Since these molecules are dysregulated in several malignancies, they represent reliable biomarkers in various contexts. In this study, the application of several Feature Selection methods uncovers a panel of 20 microRNAs, of which hsa-mir-337, hsa-mir-378c, and hsa-mir-483 are still poorly investigated in the context of Breast Cancer. This signature is capable of discriminating between healthy and tumoral samples, showing better classification performance when compared with differentially expressed microRNAs. Furthermore, a network-based centrality analysis on the gene targets of these transcripts highlighted CDC25, TPX2, KIF18B, CDCA3, TGFBR2, CAV1, TNS1, and LHFPL6 as key dysregulated genes. This study provides in silico hypotheses as new insights for future in vivo or in vitro studies uncovering the role of these putative diagnostic biomarkers in Breast Cancer.

bioinformatics↗

bioRxiv2023

MOViDA: Multi-Omics Visible Drug Activity Prediction with a Biologically Informed Neural Network Model

Drug discovery is a challenging task, characterized by a protracted period of time between initial development and market release, with a high rate of attrition at each stage. Computational virtual screening, powered by machine learning algorithms, has emerged as a promising approach for predicting therapeutic efficacy. However, the complex relationships between features learned by these algorithms can be challenging to decipher. We have devised a neural network model for the prediction of drug sensitivity, which employs a biologically-informed visible neural network (VNN), enabling a greater level of interpretability. The trained model can be scrutinized to investigate the biological pathways that play a fundamental role in prediction, as well as the chemical properties of drugs that influence sensitivity. The model leverages multi-omics data obtained from diverse tumor tissue sources and molecular descriptors that encode drug properties. We have extended the model to predict drug synergy, resulting in favorable outcomes while retaining interpretability. Given the often imbalanced nature of publicly available drug screening datasets, our model demonstrates superior performance compared to state-of-the-art visible machine learning algorithms.

bioinformatics↗

bioRxiv2021

massiveGenesetsTest: a web tool to run enrichment analysis.

MotivationInferring biological phenotypes from genomic data and sample clusters is a routinely task usually performed with Gene-Set Enrichment Analysis (GSEA), a tool that queries gene-profiles. In previous work, we scrutinized the approach based on Mann-Witney for Gene-Sets test. We highlighted the Mann-Witney test-statistics sensitivity to uncover weak signals and the drastic decreasing of time complexity. ResultsWe propose web implementation of the Gene-sets testing based on the Mann-Witney procedure. The test-procedure has reshaped to decrease the computational expense, now about tens of seconds, even if a large collection of gene-sets queries the same gene-profile. The probabilistic interpretation of the normalized test-statistic has been investigated to a better understanding of the enrichment results. A novel prioritization method across enrichment-scores, gene-set dimensions, and p-values, draws attention to relevant gene-sets. The web tool provides both tabular and graphical enrichment results. A complimentary R function allows integrating the enrichment procedure in a complex context. Contactpagnotta@unisannio.it Supplementary informationExample data and guidelines are included in the supporting material of the web-site.

bioinformatics↗

bioRxiv2020

Deep learning approaches predict non-coding RNA functions from only raw sequence data

Non-coding RNAs (ncRNAs) are small non-coding sequences involved in gene regulation in many biological processes and diseases. The lack of a complete comprehension of their biological functionality, especially in a genome-wide scenario, has demanded new computational approaches to annotate their roles. It is widely known that secondary structure is determinant to know RNA function and machine learning based approaches have been successfully proven to predict RNA function from secondary structure information. Here we show that RNA function can be predicted with good accuracy from raw sequence information without the necessity of computing secondary structure features which is computationally expensive. This finding appears to go against the dogma of secondary structure being a key determinant of function in RNA. Compared to recent secondary structure based methods, the proposed solution is more robust to sequence boundary noise and reduces drastically the computational cost allowing for large data volume annotations. Scripts and datasets to reproduce the results of experiments proposed in this study are available at: https://github.com/bioinformatics-sannio/ncrna-deep

bioinformatics↗

bioRxiv2019

A MAP of tumor-host interactions in glioma at single cell resolution

Single-cell RNA sequencing is the reference technique to characterize the heterogeneity of tumor microenvironment and can be efficiently used to discover cross-talk mechanisms between immune cells and cancer cells. We present a novel method, single cell Tumor-Host Interaction tool (scTHI), to identify significantly activated ligand-receptor interactions across clusters of cells from single-cell RNA sequencing data. We apply our approach to uncover the ligand-receptor interactions in glioma using six publicly available human glioma datasets encompassing 71 patients. We provide a comprehensive map of the signalling mechanisms between malignant cells and non-malignant cells in glioma uncovering potential novel therapeutic targets.

bioinformatics↗

bioRxiv2017

RGBM: Regularized Gradient Boosting Machines For The Identification of Transcriptional Regulators Of Discrete Glioma Subtypes

The transcription factors (TF) which regulate gene expressions are key determinants of cellular phenotypes. Reconstructing large-scale genome-wide networks which capture the influence of TFs on target genes are essential for understanding and accurate modelling of living cells. We propose RGBM: a gene regulatory network (GRN) inference algorithm, which can handle data from heterogeneous information sources including dynamic time-series, gene knockout, gene knockdown, DNA microarrays and RNA-Seq expression profiles. RGBM allows to use an a priori mechanistic of active biding network consisting of TFs and corresponding target genes. RGBM is evaluated on the DREAM challenge datasets where it surpasses the winners of the competitions and other established methods for two evaluation metrics by about 10-15%.\n\nWe use RGBM to identify the main regulators of the molecular subtypes of brain tumors. Our analysis reveals the identity and corresponding biological activities of the master regulators driving transformation of the G-CIMP-high into the G-CIMP-low subtype of glioma and PA-like into LGm6-GBM, thus, providing a clue to the yet undetermined nature of the transcriptional events driving the evolution among these novel glioma subtypes.\n\nRGBM is available for download on CRAN at https://cran.rproject.org/web/packages/RGBM/index.html

bioinformatics↗