Single-cell RNA-sequencing (scRNA-seq) allows for obtaining genomic and transcriptomic profiles of individual cells. That data makes it possible to characterize tissues at the cell level. In this context, one of the main analyses exploiting scRNAseq data is identifying the cell types within tissue to estimate the quantitative composition of cell populations. Due to the massive amount of available scRNA-seq data, automatic classification approaches for cell typing, based on the most recent deep learning technology, are needed. Here, we present the gene ontology-driven wide and deep learning (GOWDL) model for classifying cell types in several tissues. GOWDL implements a hybrid architecture that considers the functional annotations found in Gene Ontology and the marker genes typical of specific cell types. We performed cross-validation and independent external testing, comparing our algorithm with twelve other state-of-the-art predictors. Classification scores demonstrated that GOWDL reached the best results over five different tissues, except for recall, where we got about 92% vs 97% of the best tool. Finally, we presented a case study on classifying immune cell populations in breast cancer using a hierarchical approach based on GOWDL.

Breast Cancer (BC) is the most common cancer among women worldwide and is characterized by intra- and inter-tumor heterogeneity that strongly contributes towards its poor prognosis. The Estrogen Receptor (ER), Progesterone Receptor (PR), Human Epidermal Growth Factor Receptor 2 (HER2), and Ki67 antigen are the most examined markers depicting BC heterogeneity and have been shown to have a strong impact on BC prognosis. Radiomics can noninvasively predict BC heterogeneity through the quantitative evaluation of medical images, such as Magnetic Resonance Imaging (MRI), which has become increasingly important in the detection and characterization of BC. However, the lack of comprehensive BC datasets in terms of molecular outcomes and MRI modalities, and the absence of a general methodology to build and compare feature selection approaches and predictive models, limit the routine use of radiomics in the BC clinical practice. In this work, a new radiomic approach based on a two-step feature selection process was proposed to build predictors for ER, PR, HER2, and Ki67 markers. An in-house dataset was used, containing 92 multiparametric MRIs of patients with histologically proven BC and all four relevant biomarkers available. Thousands of radiomic features were extracted from post-contrast and subtracted Dynamic Contrast-Enanched (DCE) MRI images, Apparent Diffusion Coefficient (ADC) maps, and T2-weighted (T2) images. The two-step feature selection approach was used to identify significant radiomic features properly and then to build the final prediction models. They showed remarkable results in terms of F1-score for all the biomarkers: 84%, 63%, 90%, and 72% for ER, HER2, Ki67, and PR, respectively. When possible, the models were validated on the TCGA/TCIA Breast Cancer dataset, returning promising results (F1-score = 88% for the ER+/ER− classification task). The developed approach efficiently characterized BC heterogeneity according to the examined molecular biomarkers.

Many biological systems are characterised by biological entities, as well as their relationships. These interaction networks can be modelled as graphs, with nodes representing bio-entities, such as molecules, and edges representing relations among them, such as interactions. Due to the current availability of a huge amount of biological data, it is very important to consider in silico analysis methods based on, for example, machine learning, that could take advantage of the inner graph structure of the data in order to improve the quality of the results. In this scenario, graph neural networks (GNNs) are recent computational approaches that directly deal with graph-structured data. In this paper, we present a GNN network for the analysis of siRNA-mRNA interaction networks. siRNAs, in fact, are small RNA molecules that are able to bind to target genes and silence them. These events make siRNAs key molecules as RNA interference agents in many biological interaction networks related to severe diseases such as cancer. In particular, our GNN approach allows for the prediction of the siRNA efficacy, which measures the siRNA's ability to bind and silence a gene target. Tested on benchmark datasets, our proposed method overcomes other machine learning algorithms, including the state-of-the-art predictor based on the convolutional neural network, reaching a Pearson correlation coefficient of approximately 73.6%. Finally, we proposed a case study where the efficacy of a set of siRNAs is predicted for a gene of interest. To the best of our knowledge, GNNs were used for the first time in this scenario.

The incidence of Head and Neck Squamous Cell Carcinoma (HNSCC) has been growing in the last few decades. Its diagnosis is usually performed through clinical evaluation and analyzing radiological images, then confirmed by histopathological examination, an invasive and time-consuming operation. The recent advances in the artificial intelligence field are leading to interesting results in the early diagnosis, personalized treatment and monitoring of HNSCC only by analyzing radiological images, without performing a tissue biopsy. The large amount of radiological images and the increasing interest in radiomics approaches can help to develop machine learning (ML) methods to support diagnosis. In this work, we propose an ML method based on the use of radiomics features, extracted from CT and PET images, to classify the disease in terms of pN-Stage, pT-Stage and Overall Stage. After the extraction of radiomics features, a selection step is performed to remove dataset redundancy. Finally, ML methods are employed to complete the classification task. Our pipeline is applied on the "Head-Neck-PET-CT" TCIA open-source dataset, considering a cohort of 201 patients from four different institutions. An AUC of 97%, 83% and 93% in terms of pN-Stage, pT-Stage and Overall Stage classification, respectively, is achieved. The obtained results are promising, showing the potential efficiency of the use of radiomics approaches in staging classification

Recent advances in single-cell RNA-sequencing in order to study cells in biology, and the increasing amount of data available, led to the development of algorithms for analyzing single cells from gene expression data. In this work, we propose an artificial intelligence architecture that classifies cell types of human tissue. This architecture combines a deep learning model based on the convolutional neural network (CNN) with a wide model. The classification model integrates the concept of functional genes neighbourhood, based on Gene Ontology, in the CNN model (deep part) and the information on biologically relevant marker genes for each cell type in the underlying human tissue (wide part). This approach leads to a gene ontology-driven wide and deep learning model. We tested the proposed architecture with seven human tissue datasets and compared achieved results against three reference literature algorithms. Although the cell-type classification problem is heavily data-dependent, our model performed equal or better than the other models within each tissue.

In consideration of the increasing prevalence of COVID-19 cases in several countries and the resulting demand for unbiased sequencing approaches, we performed a direct RNA sequencing (direct RNA seq.) experiment using critical oropharyngeal swab samples collected from Italian patients infected with SARS-CoV-2 from the Palermo region in Sicily. Here, we identified the sequences SARS-CoV-2 directly in RNA extracted from critical samples using the Oxford Nanopore MinION technology without prior cDNA retrotranscription. Using an appropriate bioinformatics pipeline, we could identify mutations in the nucleocapsid (N) gene, which have been reported previously in studies conducted in other countries. In conclusion, to the best of our knowledge, the technique used in this study has not been used for SARS-CoV-2 detection previously owing to the difficulties in the extraction of RNA of sufficient quantity and quality from routine oropharyngeal swabs. Despite these limitations, this approach provides the advantages of true native RNA sequencing and does not include amplification steps that could introduce systematic errors. This study can provide novel information relevant to the current strategies adopted in SARS-CoV-2 next-generation sequencing.

Cytochromes P450 (CYP) are enzymes responsible for the biotransformation of most endogenous and exogenous agents. The expression of each CYP is influenced by a unique combination of mechanisms and factors including genetic polymorphisms, induction by xenobiotics, and regulation by cytokines and hormones. In recent years, Ciona robusta, one of the closest living relatives of vertebrates, has become a model in various fields of biology, in particular for studying inflammatory response. Using an in vivo LPS exposure strategy, next-generation sequencing (NGS) and qRT-PCR combined with bioinformatics and in silico analyses, compared whole pharynx transcripts from naïve and LPS-exposed C. robusta, and we provide the first view of cytochrome genes expression and miRNA regulation in the inflammatory response induced by LPS in a hematopoietic organ. In C. robusta, cytochromes belonging to 2B,2C, 2J, 2U, 4B and 4F subfamilies were deregulated and miRNA network interactions suggest that different conserved and species-specific miRNAs are involved in post-transcriptional regulation of cytochrome genes and that there could be an interplay between specific miRNAs regulating both inflammation and cytochrome molecules in the inflammatory response in C. robusta.

DNA sequences are the basic data type that is processed to perform a generic study of biological data analysis. One key component of the biological analysis is represented by sequence classification, a methodology that is widely used to analyze sequential data of different nature. However, its application to DNA sequences requires a proper representation of such sequences, which is still an open research problem. Machine Learning (ML) methodologies have given a fundamental contribution to the solution of the problem. Among them, recently, also Deep Neural Network (DNN) models have shown strongly encouraging results. In this chapter, we deal with specific classification problems related to two biological scenarios: (A) metagenomics and (B) chromatin organization. The investigations have been carried out by considering DNA sequences as input data for the classification methodologies. In particular, we study and test the efficacy of (1) different DNA sequence representations and (2) several Deep Learning (DL) architectures that process sequences for the solution of the related supervised classification problems. Although developed for specific classification tasks, we think that such architectures could be served as a suggestion for developing other DNN models that process the same kind of input.

Tumors undergo dynamic immunoediting as part of a process that balances immunologic sensing of emerging neoantigens and evasion from immune responses. Tumor-infiltrating lymphocytes (TIL) comprise heterogeneous subsets of peripheral T cells characterized by diverse functional differentiation states and dependence on T-cell receptor (TCR) specificity gained through recombination events during their development. We hypothesized that within the tumor microenvironment (TME), an antigenic milieu and immunologic interface, tumor-infiltrating peripheral T cells could reexpress key elements of the TCR recombination machinery, namely, Rag1 and Rag2 recombinases and Tdt polymerase, as a potential mechanism involved in the revision of TCR specificity. Using two syngeneic invasive breast cancer transplantable models, 4T1 and TS/A, we observed that Rag1, Rag2, and Dntt in situ mRNA expression characterized rare tumor-infiltrating T cells. In situ expression of the transcripts was increased in coisogenic Mlh1-deficient tumors, characterized by genomic overinstability, and was also modulated by PD-1 immune-checkpoint blockade. Through immunolocalization and mRNA hybridization analyses, we detected the presence of rare TDT+RAG1/2+ cells populating primary tumors and draining lymph nodes in human invasive breast cancer. Analysis of harmonized single-cell RNA-sequencing data sets of human cancers identified a very small fraction of tumor-associated T cells, characterized by the expression of recombination/revision machinery transcripts, which on pseudotemporal ordering corresponded to differentiated effector T cells. We offer thought-provoking evidence of a TIL microniche marked by rare transcripts involved in TCR shaping.

The 2,2'4,4'-tetrabromodiphenyl ether (PBDE-47) is one of the most prominent PBDE congeners detected in the environment and in animal and human tissues. Animal model experiments suggested the occurrence of PBDE-induced immunotoxicity leading to different outcomes and recently we demonstrated that this substance can impair macrophage and basophil activities. In this manuscript, we decided to further examine the effects induced by PBDE-47 treatment on innate immune response by looking at the intracellular expression profile of miRNAs as well as the biogenesis, cargo content and activity of human M(LPS) macrophage cell-derived small extracellular vesicles (sEVs). Microarray and in silico analysis demonstrated that PBDE-47 can induce some epigenetic effects in M(LPS) THP-1 cells modulating the expression of a set of intracellular miRNAs involved in biological pathways regulating the expression of estrogen-mediated signaling and immune responses with particular reference to M1/M2 differentiation. In addition to the cell-intrinsic modulation of intracellular miRNAs, we demonstrated that PBDE-47 could also interfere with the biogenesis of sEVs increasing their number and selecting a de novo population of sEVs. Moreover, PBDE-47 induced the overload of specific immune relatedmiRNAs in PBDE-47 derived sEVs. Finally, culture experiments with naïve M(LPS) macrophages demonstrated that purified PBDE-47 derived sEVs can modulate macrophage immune response exacerbating the LPSinduced pro-inflammatory response inducing the overexpression of the IL-6 and the MMP9 genes. Data from this study demonstrated that PBDE-47 can perturb the innate immune response at different levels modulating the intracellular expression of miRNAs but also interfering with the biogenesis, cargo content and functional activity of M(LPS) macrophage cellderived sEVs.

The transforming growth factor-B (TGF-B) family of cytokines performs a multifunctional signaling, which is integrated and coordinated in a signaling network that involves other pathways, such as Wintless, Forkhead box-O (FOXO) and Hedgehog and regulates pivotal functions related to cell fate in all tissues. In the hematopoietic system, TGF-B signaling controls a wide spectrum of biological processes, from immune system homeostasis to the quiescence and self-renewal of hematopoietic stem cells (HSCs). Recently an important role in post-transcription regulation has been attributed to two type of ncRNAs: microRNAs and pseudogenes. Ciona robusta, due to its philogenetic position close to vertebrates, is an excellent model to investigate mechanisms of post-transcriptional regulation evolutionarily highly conserved in immune homeostasis. The combined use of NGS and bioinformatic analyses suggests that in the pharynx, the hematopoietic organ of Ciona robusta, the Tgf-B, Wnt, Hedgehog and FoxO pathways are involved in tissue homeostasis, as they are in human. Furthermore, ceRNA network interactions and 3?UTR elements analyses of Tgf-B, Wnt, Hedgehog and FoxO pathways genes suggest that different miRNAs conserved (cin-let-7d, cin-mir-92c, cin-mir-153), species-specific (cin-mir-4187, cin-mir-4011a, cin-mir-4056, cin-mir-4150, cin-mir-4189, cin-mir-4053, cin-mir-4016, cin-mir-4075), pseudogenes (ENSCING00000011392, ENSCING00000018651, ENSCING00000007698) and mRNA 3'UTR elements are involved in post-transcriptional regulation in an integrated way in C. robusta

Studio sulla classificazione di dati radiomici relativi allo stato tumorale in pazienti affetti da carcinoma della prostata, per mezzo di variational autoencoder

Background Non-coding RNAs include different classes of molecules with regulatory functions. The most studied are microRNAs (miRNAs) that act directly inhibiting mRNA expression or protein translation through the interaction with a miRNAs-response element. Other RNA molecules participate in the complex network of gene regulation. They behave as competitive endogenous RNA (ceRNA), acting as natural miRNA sponges to inhibit miRNA functions and modulate the expression of RNA messenger (mRNA). It became evident that understanding the ceRNA-miRNA-mRNA crosstalk would increase the functional information across the transcriptome, contributing to identify new potential biomarkers for translational medicine. Results We present miRTissue ce, an improvement of our original miRTissue web service. By introducing a novel computational pipeline, miRTissue ce provides an easy way to search for ceRNA interactions in several cancer tissue types. Moreover it extends the functionalities of previous miRTissue release about miRNA-target interaction in order to provide a complete insight about miRNA mediated regulation processes. miRTissue ce is freely available at http://tblab.pa.icar.cnr.it/mirtissue.html. Conclusions The study of ceRNA networks and its dynamics in cancer tissue could be applied in many fields of translational biology, as the investigation of new cancer biomarker, both diagnostic and prognostic, and also in the investigation of new therapeutic strategies of intervention. In this scenario, miRTissue ce can offer a powerful instrument for the analysis and characterization of ceRNA-ceRNA interactions in different tissue types, representing a fundamental step in order to understand more complex regulation mechanisms.

The 16th Annual Meeting of the Bioinformatics Italian Society was held in Palermo, Italy, on June 26-28, 2019. More than 80 scientific contributions were presented, including 4 keynote lectures, 31 oral communications and 49 posters. Also, three workshops were organised before and during the meeting. Full papers from some of the works presented in Palermo were submitted for this Supplement of BMC Bioinformatics. Here, we provide an overview of meeting aims and scope. We also shortly introduce selected papers that have been accepted for publication in this Supplement, for a complete presentation of the outcomes of the meeting.

In this paper, we explore the interaction between miRNA and deregulated proteins in some pathologies. Assuming that miRNA can influence mRNA and consequently the proteins regulation, we explore this connection by using an interaction matrix derived from miRNA-target data and PPI network interactions. From this interaction matrix and the set of deregulated proteins, we search for the miRNA subset that influences the deregulated proteins with a minimum impact on the not deregulated ones. This regulation problem can be formulated as a complex optimization problem. In this paper, we have tried to solve it by using the Genetic Algorithm Heuristic. As the main result, we have found a set of miRNA that is known to be involved in disease development.

Background: In silico experiments, with the aid of computer simulation, speed up the process of in vitro or in vivo experiments. Cancer therapy design is often based on signalling pathway. MicroRNAs (miRNA) are small non-coding RNA molecules. In several kinds of diseases, including cancer, hepatitis and cardiovascular diseases, they are often deregulated, acting as oncogenes or tumor suppressors. miRNA therapeutics is based on two main kinds of molecules injection: miRNA mimics, which consists of injection of molecules that mimic the targeted miRNA, and antagomiR, which consists of injection of molecules inhibiting the targeted miRNA. Nowadays, the research is focused on miRNA therapeutics. This paper addresses cancer related signalling pathways to investigate miRNA therapeutics. Results: In order to prove our approach, we present two different case studies: non-small cell lung cancer and melanoma. KEGG signalling pathways are modelled by a digital circuit. A logic value of 1 is linked to the expression of the corresponding gene. A logic value of 0 is linked to the absence (not expressed) gene. All possible relationships provided by a signalling pathway are modelled by logic gates. Mutations, derived according to the literature, are introduced and modelled as well. The modelling approach and analysis are widely discussed within the paper. MiRNA therapeutics is investigated by the digital circuit analysis. The most effective miRNA and combination of miRNAs, in terms of reduction of pathogenic conditions, are obtained. A discussion of obtained results in comparison with literature data is provided. Results are confirmed by existing data. Conclusions: The proposed study is based on drug discovery and miRNA therapeutics and uses a digital circuit simulation of a cancer pathway. Using this simulation, the most effective combination of drugs and miRNAs for mutated cancer therapy design are obtained and these results were validated by the literature. The proposed modelling and analysis approach can be applied to each human disease, starting from the corresponding signalling pathway.

Data mining is the set of computational techniques and methodologies aimed to extract knowledge from a large amount of data, by using sophisticated data analysis tools to highlight information structure underlying large data sets. Machine learning methods represent one of these tools, allowing, not only data management but also analysis and prediction operations. Supervised learning, a kind of machine learning methodology, uses input data and products outputs of two type: qualitative and quantitative, respectively describing data classes and predicting data trends. Classification task provides qualitative responses whereas prediction or regression task offers quantitative outputs