Harnessing deep learning to forecast local microclimate using global climate data.

Microclimate is a complex non-linear phenomenon influenced by both global and local processes. Its understanding holds a pivotal role in the management of natural resources and the optimization of agricultural procedures. This phenomenon can be effectively monitored in local areas by employing models that integrate physical laws and data-driven algorithms relying on climate data and terrain conformation. Climate data can be acquired from nearby meteorological stations when available, but in their absence, global climate datasets describing 10 km-scale areas are often utilized. The present research introduces an innovative microclimate model that combines physical laws and deep learning to reproduce temperature and relative humidity variations at the meter-scale within a study area located in the Lombardian foothills. The model is exploited to perform a comparative study investigating whether employing the global climate dataset ERA5 as input reduces model's accuracy in reproducing the microclimate variations compared to using data collected by the Lombardy Regional Environment Protection Agency (ARPA) from a nearby meteorological station. The comparative analysis shows that using local meteorological data as inputs provides more accurate results for microclimate modeling. However, in situations where local data is not available, the use of global climate data remains a viable and reliable approach.

Interrogating colorectal cancer metastasis to liver: a search for clinically viable compounds and mechanistic insights in colorectal cancer Patient Derived Organoids.

BACKGROUND: Approximately 20-50% of patients presenting with localized colorectal cancer progress to stage IV metastatic disease (mCRC) following initial treatment and this is a major prognostic determinant. Here, we have interrogated a heterogeneous set of primary colorectal cancer (CRC), liver CRC metastases and adjacent liver tissue to identify molecular determinants of the colon to liver spreading. Screening Food and Drug Administration (FDA) approved drugs for their ability to interfere with an identified colon to liver metastasis signature may help filling an unmet therapeutic need. METHODS: RNA sequencing of primary colorectal cancer specimens vs adjacent liver tissue vs synchronous and asynchronous liver metastases. Pathways enrichment analyses. The Library of Integrated Network-based Cellular Signatures (LINCS)-based and Connectivity Map (CMAP)-mediated identification of FDA-approved compounds capable to interfere with a 22 gene signature from primary CRC and liver metastases. Testing the identified compounds on CRC-Patient Derived Organoid (PDO) cultures. Microscopy and Fluorescence Activated Cell Sorting (FACS) based analysis of the treated PDOs. RESULTS: We have found that liver metastases acquire features of the adjacent liver tissue while partially losing those of the primary tumors they derived from. We have identified a 22-gene signature differentially expressed among primary tumors and metastases and validated in public databases. A pharmacogenomic screening for FDA-approved compounds capable of interfering with this signature has been performed. We have validated some of the identified representative compounds in CRC-Patient Derived Organoid cultures (PDOs) and found that pentoxyfilline and, to a minor extent, dexketoprofen and desloratadine, can variably interfere with number, size and viability of the CRC -PDOs in a patient-specific way. We explored the pentoxifylline mechanism of action and found that pentoxifylline treatment attenuated the 5-FU elicited increase of ALDHhigh cells by attenuating the IL-6 mediated STAT3 (tyr705) phosphorylation. CONCLUSIONS: Pentoxifylline synergizes with 5-Fluorouracil (5-FU) in attenuating organoid formation. It does so by interfering with an IL-6-STAT3 axis leading to the emergence of chemoresistant ALDHhigh cell subpopulations in 5-FU treated PDOs. A larger cohort of CRC-PDOs will be required to validate and expand on the findings of this proof-of-concept study.

Infiltration of tumor spheroids by activated immune cells.

Recent years have seen a tremendous growth of interest in understanding the role that the adaptive immune system could play in interdicting tumor progression. In this context, it has been shown that the density of adaptive immune cells inside a solid tumor serves as a favorable prognostic marker across different types of cancer. The exact mechanisms underlying the degree of immune cell infiltration is largely unknown. Here, we quantify the temporal dynamics of the density profile of activated immune cells around a solid tumor spheroid. We propose a computational model incorporating immune cells with active, persistent movement and a proliferation rate that depends on the presence of cancer cells, and show that the model able to reproduce semi-quantitatively the experimentally measured infiltration profile. Studying the density distribution of immune cells inside a solid tumor can help us better understand immune trafficking in the tumor micro-environment, hopefully leading towards novel immunotherapeutic strategies.

The Response of Triple-Negative Breast Cancer to Neoadjuvant Chemotherapy and the Epithelial-Mesenchymal Transition.

It would be highly desirable to find prognostic and predictive markers for triple-negative breast cancer (TNBC), a strongly heterogeneous and invasive breast cancer subtype often characterized by a high recurrence rate and a poor outcome. Here, we investigated the prognostic and predictive capabilities of ARIADNE, a recently developed transcriptomic test focusing on the epithelial-mesenchymal transition. We first compared the stratification of TNBC patients obtained by ARIADNE with that based on other common pathological indicators, such as grade, stage and nodal status, and found that ARIADNE was more effective than the other methods in dividing patients into groups with different disease-free survival statistics. Next, we considered the response to neoadjuvant chemotherapy and found that the classification provided by ARIADNE led to statistically significant differences in the rates of pathological complete response within the groups.

Quantitative analysis of disease-related metabolic dysregulation of human microbiota.

The metabolic activity of all the micro-organism composing the human microbiome interacts with the host metabolism contributing to human health and disease in a way that is not fully understood. Here, we introduce STELLA, a computational method to derive the spectrum of metabolites associated with the microbiome of an individual. STELLA integrates known information on metabolic pathways associated with each bacterial species and extracts from these the list of metabolic products of each singular reaction by means of automatic text analysis. By comparing the result obtained on a single subject with the metabolic profile data of a control set of healthy subjects, we are able to identify individual metabolic alterations. To illustrate the method, we present applications to autism spectrum disorder and multiple sclerosis.

A pipeline for monitoring water pollution: The example of heavy metals in Lombardy waters.

Time-dependent geolocalized analysis of pollution data allows to better understand their dynamics over time and could suggest strategies to restore a good ecological status of contaminated area. This research analyzes concentrations of pollutants in surface waters and groundwater monitored by the Regional Environment Protection Agency of Lombardy from 2017 to 2020. Lombardy is one of the richest and populous region of Europe, providing an interesting example of the impact of environmental pollutants due to anthropogenic and industrial activities, not only for Italy but also for all Europe. Results show that groundwater displays more sites with heavy metals above the legal limit with respect to surface waters, including As, Ni, Cr and Zn. Furthermore, the spatio-temporal analysis of the data clearly shows that the introduction of more restrictive laws is a proper policy to improve the ecological status of the water.

Artificial intelligence in breast cancer diagnostics.

Since breast cancer deaths are mainly due to metastasis, predicting the risk that a primary tumor will develop metastasis after a first diagnosis is a central issue that could be addressed by artificial intelligence. To overcome the problem posed by limited availability of standardized datasets, algorithms should include biological insight.

Classification of triple negative breast cancer by epithelial mesenchymal transition and the tumor immune microenvironment.

Triple-negative breast cancer (TNBC) accounts for about 15-20% of all breast cancers and differs from other invasive breast cancer types because it grows and spreads rapidly, it has limited treatment options and typically worse prognosis. Since TNBC does not express estrogen or progesterone receptors and little or no human epidermal growth factor receptor (HER2) proteins are present, hormone therapy and drugs targeting HER2 are not helpful, leaving chemotherapy only as the main systemic treatment option. In this context, it would be important to find molecular signatures able to stratify patients into high and low risk groups. This would allow oncologists to suggest the best therapeutic strategy in a personalized way, avoiding unnecessary toxicity and reducing the high costs of treatment. Here we compare two independent patient stratification strategies for TNBC based on gene expression data: The first is focusing on the epithelial mesenchymal transition (EMT) and the second on the tumor immune microenvironment. Our results show that the two stratification strategies are not directly related, suggesting that the aggressiveness of the tumor can be due to a multitude of unrelated factors. In particular, the EMT stratification is able to identify a high-risk population with high immune markers that is, however, not properly classified by the tumor immune microenvironment based strategy.

Universal density of low-frequency states in silica glass at finite temperatures.

The theoretical understanding of the low-frequency modes in amorphous solids at finite temperature is still incomplete. The study of the relevant modes is obscured by the dressing of interparticle forces by collision-induced momentum transfer that is unavoidable at finite temperatures. Recently, it was proposed that low-frequency modes of vibrations around the thermally averaged configurations deserve special attention. In simple model glasses with bare binary interactions, these included quasilocalized modes whose density of states appears to be universal, depending on the frequencies as D(ω)∼ω^{4}, in agreement with the similar law that is obtained with bare forces at zero temperature. In this paper, we report investigations of a model of silica glass at finite temperature; here the bare forces include binary and ternary interactions. Nevertheless, we can establish the validity of the universal law of the density of quasilocalized modes also in this richer and more realistic model glass.

Predicting the failure of two-dimensional silica glasses.

Being able to predict the failure of materials based on structural information is a fundamental issue with enormous practical and industrial relevance for the monitoring of devices and components. Thanks to recent advances in deep learning, accurate failure predictions are becoming possible even for strongly disordered solids, but the sheer number of parameters used in the process renders a physical interpretation of the results impossible. Here we address this issue and use machine learning methods to predict the failure of simulated two dimensional silica glasses from their initial undeformed structure. We then exploit Gradient-weighted Class Activation Mapping (Grad-CAM) to build attention maps associated with the predictions, and we demonstrate that these maps are amenable to physical interpretation in terms of topological defects and local potential energies. We show that our predictions can be transferred to samples with different shape or size than those used in training, as well as to experimental images. Our strategy illustrates how artificial neural networks trained with numerical simulation results can provide interpretable predictions of the behavior of experimentally measured structures.

Evidence of a SARS-CoV-2 double Spike mutation D614G/S939F potentially affecting immune response of infected subjects.

OBJECTIVES: Despite extensive efforts to monitor the diffusion of COVID-19, the actual wave of infection is worldwide characterized by the presence of emerging SARS-CoV-2 variants. The present study aims to describe the presence of yet undiscovered SARS-CoV-2 variants in Italy. METHODS: Next Generation Sequencing was performed on 16 respiratory samples from occasionally employed within the Bangladeshi community present in Ostia and Fiumicino towns. Computational strategy was used to identify all potential epitopes for reference and mutated Spike proteins. A simulation of proteasome activity and the identification of possible cleavage sites along the protein guided to a combined score involving binding affinity, peptide stability and T-cell propensity. RESULTS: Retrospective sequencing analysis revealed a double Spike D614G/S939F mutation in COVID-19 positive subjects present in Ostia while D614G mutation was evidenced in those based in Fiumicino. Unlike D614G, S939F mutation affects immune response by the slight but significant modulation of T-cell propensity and the selective enrichment of potential binding epitopes for some HLA alleles. CONCLUSION: Collectively, our findings mirror further the importance of deep sequencing of SARS-CoV-2 genome as a unique approach to monitor the appearance of specific mutations as for those herein reported for Spike protein. This might have implications on both the type of immune response triggered by the viral infection and the severity of the related illness.

Immune Profile of SARS-CoV-2 Variants of Concern.

The spread of the current Sars-Cov-2 pandemics leads to the development of mutations that are constantly monitored because they could affect the efficacy of vaccines. Three recently identified mutated strains, known as variants of concern, are rapidly spreading worldwide. Here, we study possible effects of these mutations on the immune response to Sars-Cov-2 infection using NetTepi a computational method based on artificial neural networks that considers binding and stability of peptides obtained by proteasome degradation for widely represented HLA class I alleles present in human populations as well as the T-cell propensity of viral peptides that measures their immune response. Our results show variations in the number of potential highly ranked peptides ranging between 0 and 20% depending on the specific HLA allele. The results can be useful to design more specific vaccines.

Synergistic effects of contaminants in Lombardy waters.

Quantifying synergistic environmental effects in water contamination is still an open issue. Here, we have analyzed geolocalized data of pollutants recorded in 2018 in surface and groundwater of Lombardy, one of the areas with the highest agricultural production rates, not only in Italy, but also in Europe. Both herbicides and insecticides are present at concentration levels above the legal limit, mainly in surface waters. Geolocalized analysis allows us to identify interesting areas particularly affected by a combination of multiple pesticides. We thus investigated possible synergistic effects of these compounds on the environment, using the alga C. reinhardtii as a biosensor. Our results show that exposure for 7 days to four compounds, that we found present together at high concentration in surface waters, was able to induce a stress in the algae, as indicated by the presence of palmelloids. Our work results in a pipeline that could easily be exported to monitor other territories in Italy and abroad.

Role of body temperature variations in bat immune response to viral infections.

The ability of bats to coexist with viruses without being harmed is an interesting issue that is still under investigation. Here we use a mathematical model to show that the pattern of body temperature variations observed in bats between day and night is responsible for their ability to keep viruses in check. From the dynamical systems point of view, our model displays an intriguing quasi-periodic behaviour that might be relevant in making the system robust by avoiding viral escape due to perturbations in the body temperature cycle.

Classification of triple-negative breast cancers through a Boolean network model of the epithelial-mesenchymal transition.

Predicting the metastasis risk in patients with a primary breast cancer tumor is of fundamental importance to decide the best therapeutic strategy in the framework of personalized medicine. Here, we present ARIADNE, a general algorithmic strategy to assess the risk of metastasis from transcriptomic data of patients with triple-negative breast cancer, a subtype of breast cancer with poorer prognosis with respect to the other subtypes. ARIADNE identifies hybrid epithelial/mesenchymal phenotypes by mapping gene expression data into the states of a Boolean network model of the epithelial-mesenchymal pathway. Using this mapping, it is possible to stratify patients according to their prognosis, as we show by validating the strategy with three independent cohorts of triple-negative breast cancer patients. Our strategy provides a prognostic tool that could be applied to other biologically relevant pathways, in order to estimate the metastatic risk for other breast cancer subtypes or other tumor types. A record of this paper's transparent peer review process is included in the supplemental information.

Information Optimized Multilayer Network Representation of High Density Electroencephalogram Recordings.

High-density electroencephalography (hd-EEG) provides an accessible indirect method to record spatio-temporal brain activity with potential for disease diagnosis and monitoring. Due to their highly multidimensional nature, extracting useful information from hd-EEG recordings is a complex task. Network representations have been shown to provide an intuitive picture of the spatial connectivity underlying an electroencephalogram recording, although some information is lost in the projection. Here, we propose a method to construct multilayer network representations of hd-EEG recordings that maximize their information content and test it on sleep data recorded in individuals with mental health issues. We perform a series of statistical measurements on the multilayer networks obtained from patients and control subjects and detect significant differences between the groups in clustering coefficient, betwenness centrality, average shortest path length and parieto occipital edge presence. In particular, patients with a mood disorder display a increased edge presence in the parieto-occipital region with respect to healthy control subjects, indicating a highly correlated electrical activity in that region of the brain. We also show that multilayer networks at constant edge density perform better, since most network properties are correlated with the edge density itself which can act as a confounding factor. Our results show that it is possible to stratify patients through statistical measurements on a multilayer network representation of hd-EEG recordings. The analysis reveals that individuals with mental health issues display strongly correlated signals in the parieto-occipital region. Our methodology could be useful as a visualization and analysis tool for hd-EEG recordings in a variety of pathological conditions.

Oral mucositis: the hidden side of cancer therapy.

Inflammation response of epithelial mucosa to chemo- radiotherapy cytotoxic effects leads to mucositis, a painful side effect of antineoplastic treatments. About 40% of the patients treated with chemotherapy develop mucositis; this percentage rises to about 90% for head and neck cancer patients (HNC) treated with both chemo- and radiotherapy. 19% of the latter will be hospitalized and will experience a delay in antineoplastic treatment for high-grade mucositis management, resulting in a reduction of the quality of life, a worse prognosis and an increase in patient management costs. Currently, several interventions and prevention guidelines are available, but their effectiveness is uncertain. This review comprehensively describes mucositis, debating the impact of standard chemo-radiotherapy and targeted therapy on mucositis development and pointing out the limits and the benefits of current mucositis treatment strategies and assessment guidelines. Moreover, the review critically examines the feasibility of the existing biomarkers to predict patient risk of developing oral mucositis and their role in early diagnosis. Despite the expression levels of some proteins involved in the inflammation response, such as TNF-α or IL-1ß, partially correlate with mucositis process, their presence does not exclude others mucositis-independent inflammation events. This strongly suggests the need to discover biomarkers that specifically feature mucositis process development. Non-coding RNAs might hold this potential.

Estimating the Binding of Sars-CoV-2 Peptides to HLA Class I in Human Subpopulations Using Artificial Neural Networks.

Epidemiological studies show that SARS-CoV-2 infection leads to severe symptoms only in a fraction of patients, but the determinants of individual susceptibility to the virus are still unknown. The major histocompatibility complex (MHC) class I exposes viral peptides in all nucleated cells and is involved in the susceptibility to many human diseases. Here, we use artificial neural networks to analyze the binding of SARS-CoV-2 peptides with polymorphic human MHC class I molecules. In this way, we identify two sets of haplotypes present in specific human populations: the first displays weak binding with SARS-CoV-2 peptides, while the second shows strong binding and T cell propensity. Our work offers a useful support to identify the individual susceptibility to COVID-19 and illustrates a mechanism underlying variations in the immune response to SARS-CoV-2. A record of this paper's transparent peer review process is included in the Supplemental Information.

Universal Low-Frequency Vibrational Modes in Silica Glasses.

It was recently shown that different simple models of glass formers with binary interactions define a universality class in terms of the density of states of their quasilocalized low-frequency modes. Explicitly, once the hybridization with standard Debye (extended) modes is avoided, a number of such models exhibit a universal density of states, depending on the mode frequencies as D(ω)∼ω^{4}. It is unknown, however, how wide this universality class is, and whether it also pertains to more realistic models of glass formers. To address this issue we present analysis of the quasilocalized modes in silica, a network glass that has both binary and ternary interactions. We conclude that in three dimensions silica exhibits the very same frequency dependence at low frequencies, suggesting that this universal form is a generic consequence of amorphous glassiness.

Cell-cell adhesion and 3D matrix confinement determine jamming transitions in breast cancer invasion.

Plasticity of cancer invasion and metastasis depends on the ability of cancer cells to switch between collective and single-cell dissemination, controlled by cadherin-mediated cell-cell junctions. In clinical samples, E-cadherin-expressing and -deficient tumours both invade collectively and metastasize equally, implicating additional mechanisms controlling cell-cell cooperation and individualization. Here, using spatially defined organotypic culture, intravital microscopy of mammary tumours in mice and in silico modelling, we identify cell density regulation by three-dimensional tissue boundaries to physically control collective movement irrespective of the composition and stability of cell-cell junctions. Deregulation of adherens junctions by downregulation of E-cadherin and p120-catenin resulted in a transition from coordinated to uncoordinated collective movement along extracellular boundaries, whereas single-cell escape depended on locally free tissue space. These results indicate that cadherins and extracellular matrix confinement cooperate to determine unjamming transitions and stepwise epithelial fluidization towards, ultimately, cell individualization.