Tumor-educated platelets (TEPs) are circulating blood cells implicated as central players in the systemic and local responses to tumor growth, thus altering their RNA profile. To date, some studies have shown that the TEPs transcriptome can be used for a less invasive cancer diagnosis. The objective of this study is to propose a procedure that can identify a set of key genes with diagnostic value for glioblastoma multiforme (GBM). To identify these key genes, we analyzed TEPs RNA-seq data of healthy subjects and GBM patients from two different public datasets (one used as the main dataset the other as test set). We performed differential expression analysis (DEA) and differential co-expression (DCE) network analysis. Specifically, leveraging the main dataset, we first performed DEA analysis to identify differentially expressed genes (DEGs) and then used these genes to construct and analyze the differential co-expression network. From this network, we extracted centrality metrics and local clustering coefficient to identify key nodes, hence the more suitable genes for diagnostic purposes. Then we tested these key genes on the other dataset. Our findings show that genes identified by betweenness centrality exhibit superior diagnostic power compared to: DEGs, gene sets identified through other metrics, and random sets of differentially expressed genes.

Radiomics represents a specialized branch of medical imaging where quantitative features are extracted from images. Performing a classification using radiomics means solving two common problems: the imbalanced setting, and the large number of features that would increase the risk of overfitting. Moreover, since its main application and impact are in clinical field, there is the need of interpretable models for explaining their results. The aim of this study is to compare two modelling approaches: one based on a logistic regression model, known for its simplicity and interpretability, and RUSBoost, an ensemble method designed to handle class imbalance with potentially higher complexity, in order to answer the question whether higher complexity and lower interpretability are justified when dealing with radiomics data. Additionally, due to the large literature suggesting it, we analyze the impact of a feature selection step applied to these two classifiers. Test performances measured across 20 repeated splits on two datasets show how the RUSBoost approach is able to capture more detailed patterns of the data but this is highly dependent on the dataset at hand.

Global network alignment is the computational problem of determining the similarity between nodes of different networks to establish a one-to-one correspondence between them. It has important applications in the biological field, particularly for discovering similar roles between the elements of different systems or for transferring knowledge from a well-studied system to another. In this paper, we present SHELLEY, a tool that facilitates the development, testing, and combination of learning-based network alignment algorithms by providing a set of modules that allow for the recreation and combination of both representation learning methods (RLMs) and deep matching methods (DMMs). We then present a case study in which we apply this tool to a protein-protein interaction network (PPI), demonstrating how the representation phase of RLMs is crucial for model robustness against noise.The code of SHELLEY is available at: https://github.com/rickydeluca/shelley

Drug repositioning is a promising strategy to discover new therapeutic applications for existing drugs, significantly reducing the time and costs associated with traditional drug development. This study employs a network medicine approach to analyze successful cases of drug repositioning, focusing on the exploratory hypothesis that the efficacy of repositioning may be determined by functional similarity between between diseases for which the drug was originally designed and diseases for which the same drug is reused. Network medicine tools were employed to investigate the connections between disease-associated genes, proteins, and approved drugs. Biological networks, including protein-protein interactions and functional interactions networks, as well as gene- and drug-disease association data are analyzed to identify functional similarities and possible molecular connections between diseases and treatments. Using clustering techniques and topological analysis, the results reveal a suggestive overlap of involved genes and functional interactions, emphasizing the value of computational methods in accelerating drug repositioning efforts and improving understanding of drug repositioning dynamics for more efficient therapeutic interventions.

Genetic instability is one of the hallmarks of cancer, however mutations can occur for different causes and induce different effects. Mutational signatures are characteristic patterns of somatic mutations in cancer genomes, reflecting the underlying mutational processes. A mutational signature can be determined by studying the kind of mutations a patient has acquired during their life: patient stratification based on mutational signatures has become more and more useful in genomic studies, given its possible clinical implications. In this work we focused on Single Base Substitution (SBS) signatures to study a cohort of 115 metastatic melanoma patients. We inferred and identified two mutational signatures characterizing the patients. Based on these signatures we divided patients into two group: the bigger group was characterized by a signature associated with exposure to ultraviolet light, while the smaller group resulted to be mostly composed of patients which did not respond to immunotherapy (anti-PD1) and that presented a low mutational count. More importantly this second group showed a significantly worse survival outcome. The use of mutational signatures is clearly a powerful tool to identify disease sub-types that have a clinical relevance, however we believe that this topic needs further investigation focused on the characterization of patient subtypes with a multi-omics based approach.

Recent innovations and developments in molecular biology and biotechnology have made it possible to acquire and store large omics datasets. In particular genomics, transcriptomics and the study of their relationship represent the key elements to understand how genotype influences phenotype. In this work we investigate the potential of multi-layer network modeling in integrating genomic and transcriptomic data of 152 advanced non-small cell lung cancer (NSCLC) patients treated with anti-PD-(L)1 therapy. For the transcriptomic layer, we performed differential expression and differential co-expression analyses in order to identify a subset of key genes in differentiating responder patients from non-responders. Adding to the genomic-transcriptomic model other three layers related to immune, myeloid and curated immunotherapy-based literature signatures we obtained a 5-layer Patient Similarity Network. The application of Similarity Network Fusion algorithm revealed a statistically significant stratification of patients (p-value ≤ 0.05) which allows the identification of two clusters characterized by patients responding and not responding to immunotherapy.

Colorectal cancer is the third most diagnosed cancer in the world, but it has a higher mortality rate in men compared to women. However, we are not close to understanding how and why sex influences the outcome of the disease. This study focuses on mRNA expression profiles of colon cancer patients to look for molecular differences in the development of colon cancer between men and women. We used paired expression data (i.e., data collected in pairs of normal and cancer cells, by taking samples from the same individual), we identified differentially expressed genes (cancer vs normal) and computed co-expression and differential co-expression gene networks (men vs women). Doing so, we inferred the main changes and alterations happening in cancer tissues, and specifically how these changes were different among men and women. We found that the co-expression networks of women and men affected by colon cancer are quite different and we reported the genes that show the most differences in this comparison, checking if they could also be associated to sexual dimorphism or sexual hormones. Among these genes we found a interesting presence of genes associated to the Wnt signaling pathway which has been found to be regulated by estrogen and whose activation is strongly linked with colon cancer.

The possibility to computationally prioritize candidate disease genes capitalizing on existing information has led to a speedup in the discovery of new methods. Many gene discovery techniques exploit network data, like protein-protein interactions (PPIs), in order to extract knowledge from the network structure relying on several network metrics. We here present PROCONSUL, a method that builds on top of the concept of connectivity significance (CS) and exploits the idea of probabilistic exploration of the space of putative disease genes. We show that our methodology is able to outperform the state-of-the-art tool based on CS in several settings, and propose different, effective gene discovery strategies according to specific disease network properties.

Tumor-educated platelets (TEPs) are circulating blood cells with a distinct tumor-driven phenotype and act as carriers and protectors of metastases. To date, some studies have shown that the TEPs transcriptome can be used for cancer diagnostics. The objective of this study is to propose a procedure based on differential gene expression and differential gene co-expression analyses to identify a set of key genes for multi-class cancer diagnostics. To reach this aim, we analyzed RNA-seq data (57736 genes) of 130 subjects, of whom 40 patients with glioblastoma multiforme (GBM), 35 patients with pancreatic adenocarcinoma (PAAD), and 55 healthy donors (HC). We focused our analysis on the subset of differentially expressed genes (DEGs), and we used these genes to build and analyze the differential co-expression networks, identifying the hub genes. With this procedure, we obtained a restricted set of DEGs that maximize the accuracy in classifying patients according to their conditions (GBM, PAAD, or HC). Indeed, we validated our results by comparing the achieved classification accuracy with that resulting from random selections of DEGs and we obtained that genes selected by differential co-expression (DCE) network analysis have greater predictive power than any other set of differentially expressed genes, including using all of them.

A fundamental topic in network medicine is disease genes prioritization. The underlying hypothesis is that disease genes are organized as modules confined within the interactome. Here, we propose a novel algorithm called DiaBLE (DIAMOnD Background Local Expansion) which is a modified version of DIAMOnD, a successful algorithm based on the concept of connectivity significance. Instead of taking the whole interactome as the background model, DiaBLE considers as gene universe the smallest local expansion of the current seeds set at each iteration step. We show that DiaBLE significantly increases the overall DIAMOnD ranking quality of genes prioritization both in terms of cross-validation and biological consistency. Here, we focus on the two algorithms only since a comparative analysis among gene prioritization methods is beyond the scope of this study. Finally, we briefly discuss the improvement of biological insight provided by DiaBLE for two cancers (head and neck squamous cell carcinoma and kidney renal clear cell carcinoma).

This work proposes a unified framework to leverage biological information in network propagation-based gene prioritization algorithms. Preliminary results on breast cancer data show significant improvements over state-of-the-art baselines, such as the prioritization of genes that are not identified as potential candidates by interactome-based algorithms, but that appear to be involved in/or potentially related to breast cancer, according to a functional analysis based on recent literature.