Stroke Treatment Prediction Using Features Selection Methods and Machine Learning Classifiers

ObjectivesFeature selection in data sets is an important task allowing to alleviate various machine learning and data mining issues. The main objectives of a feature selection method consist on building simpler and more understandable classifier models in order to improve the data mining and processing performances. Therefore, a comparative evaluation of the Chi-square method, recursive feature elimination method, and tree-based method (using Random Forest) used on the three common machine learning methods (K-Nearest Neighbor, naïve Bayesian classifier and decision tree classifier) are performed to select the most relevant primitives from a large set of attributes. Furthermore, determining the most suitable couple (i.e., feature selection method-machine learning method) that provides the best performance is performed.Materials and methodsIn this paper, an overview of the most common feature selection techniques is first provided: the Chi-Square method, the Recursive Feature Elimination method (RFE) and the tree-based method (using Random Forest). A comparative evaluation of the improvement (brought by such feature selection methods) to the three common machine learning methods (K- Nearest Neighbor, naïve Bayesian classifier and decision tree classifier) are performed. For evaluation purposes, the following measures: micro-F1, accuracy and root mean square error are used on the stroke disease data set.ResultsThe obtained results show that the proposed approach (i.e., Tree Based Method using Random Forest, TBM-RF, decision tree classifier, DTC) provides accuracy higher than 85%, F1-score higher than 88%, thus, better than the KNN and NB using the Chi-Square, RFE and TBM-RF methods.ConclusionThis study shows that the couple - Tree Based Method using Random Forest (TBM-RF) decision tree classifier successfully and efficiently contributes to find the most relevant features and to predict and classify patient suffering of stroke disease.”     

Toward high-content screening of mitochondrial morphology and membrane potential in living cells

Mitochondria are double membrane organelles involved in various key cellular processes. Governed by dedicated protein machinery, mitochondria move and continuously fuse and divide. These “mitochondrial dynamics” are bi-directionally linked to mitochondrial and cell functional state in space and time. Due to the action of the electron transport chain (ETC), the mitochondrial inner membrane displays a inside-negative membrane potential (Δψ). The latter is considered a functional readout of mitochondrial “health” and required to sustain normal mitochondrial ATP production and mitochondrial fusion. During the last decade, live-cell microscopy strategies were developed for simultaneous quantification of Δψ and mitochondrial morphology. This revealed that ETC dysfunction, changes in Δψ and aberrations in mitochondrial structure often occur in parallel, suggesting they are linked potential targets for therapeutic intervention. Here we discuss how combining high-content and high-throughput strategies can be used for analysis of genetic and/or drug-induced effects at the level of individual organelles, cells and cell populations.This article is part of a Directed Issue entitled: Energy Metabolism Disorders and Therapies.     

Predicting CK2 beta-dependent substrates using linear patterns

CK2 is a constitutively active Ser/Thr protein kinase deregulated in cancer and other pathologies, responsible for about the 20% of the human phosphoproteome. The holoenzyme is a complex composed of two catalytic (α or α´) and two regulatory (β) subunits, with individual subunits also coexisting in the cell. In the holoenzyme, CK2β is a substrate-dependent modulator of kinase activity. Therefore, a comprehensive characterization of CK2 cellular function should firstly address which substrates are phosphorylated exclusively when CK2β is present (class-III or beta-dependent substrates). However, current experimental constrains limit this classification to a few substrates. Here, we took advantage of motif-based prediction and designed four linear patterns for predicting class-III behavior in sets of experimentally determined CK2 substrates. Integrating high-throughput substrate prediction, functional classification and network analysis, our results suggest that beta-dependent phosphorylation might exert particular regulatory roles in viral infection and biological processes/pathways like apoptosis, DNA repair and RNA metabolism. It also pointed, that human beta-dependent substrates are mainly nuclear, a few of them shuttling between nuclear and cytoplasmic compartments.     

Consolidation of proteomics data in the Cancer Proteomics database

Cancer is a class of diseases characterized by abnormal cell growth and one of the major reasons for human deaths. Proteins are involved in the molecular mechanisms leading to cancer, furthermore they are affected by anti‐cancer drugs, and protein biomarkers can be used to diagnose certain cancer types. Therefore, it is important to explore the proteomics background of cancer. In this report, we developed the Cancer Proteomics database to re‐interrogate published proteome studies investigating cancer. The database is divided in three sections related to cancer processes, cancer types, and anti‐cancer drugs. Currently, the Cancer Proteomics database contains 9778 entries of 4118 proteins extracted from 143 scientific articles covering all three sections: cell death (cancer process), prostate cancer (cancer type) and platinum‐based anti‐cancer drugs including carboplatin, cisplatin, and oxaliplatin (anti‐cancer drugs). The detailed information extracted from the literature includes basic information about the articles (e.g., PubMed ID, authors, journal name, publication year), information about the samples (type, study/reference, prognosis factor), and the proteomics workflow (Subcellular fractionation, protein, and peptide separation, mass spectrometry, quantification). Useful annotations such as hyperlinks to UniProt and PubMed were included. In addition, many filtering options were established as well as export functions. The database is freely available at http://cancerproteomics.uio.no .     

Genome mining reveals unlocked bioactive potential of marine Gram-negative bacteria

Background

Antibiotic resistance in bacteria spreads quickly, overtaking the pace at which new compounds are discovered and this emphasizes the immediate need to discover new compounds for control of infectious diseases. Terrestrial bacteria have for decades been investigated as a source of bioactive compounds leading to successful applications in pharmaceutical and biotech industries. Marine bacteria have so far not been exploited to the same extent; however, they are believed to harbor a multitude of novel bioactive chemistry. To explore this potential, genomes of 21 marine Alpha- and Gammaproteobacteria collected during the Galathea 3 expedition were sequenced and mined for natural product encoding gene clusters.

Results

Independently of genome size, bacteria of all tested genera carried a large number of clusters encoding different potential bioactivities, especially within the Vibrionaceae and Pseudoalteromonadaceae families. A very high potential was identified in pigmented pseudoalteromonads with up to 20 clusters in a single strain, mostly NRPSs and NRPS-PKS hybrids. Furthermore, regulatory elements in bioactivity-related pathways including chitin metabolism, quorum sensing and iron scavenging systems were investigated both in silico and in vitro. Genes with siderophore function were identified in 50% of the strains, however, all but one harboured the ferric-uptake-regulator gene. Genes encoding the syntethase of acylated homoserine lactones were found in Roseobacter-clade bacteria, but not in the Vibrionaceae strains and only in one Pseudoalteromonas strains. The understanding and manipulation of these elements can help in the discovery and production of new compounds never identified under regular laboratory cultivation conditions. High chitinolytic potential was demonstrated and verified for Vibrio and Pseudoalteromonas species that commonly live in close association with eukaryotic organisms in the environment. Chitin regulation by the ChiS histidine-kinase seems to be a general trait of the Vibrionaceae family, however it is absent in the Pseudomonadaceae. Hence, the degree to which chitin influences secondary metabolism in marine bacteria is not known.

Conclusions

Utilizing the rapidly developing sequencing technologies and software tools in combination with phenotypic in vitro assays, we demonstrated the high bioactive potential of marine bacteria in an efficient, straightforward manner – an approach that will facilitate natural product discovery in the future.     

A spatial genomic approach identifies time lags and historical barriers to gene flow in a rapidly fragmenting Appalachian landscape

The resolution offered by genomic data sets coupled with recently developed spatially informed analyses are allowing researchers to quantify population structure at increasingly fine temporal and spatial scales. However, both empirical research and conservation measures have been limited by questions regarding the impacts of data set size, data quality thresholds and the timescale at which barriers to gene flow become detectable. Here, we used restriction site associated DNA sequencing to generate a 2,140 single nucleotide polymorphism (SNP) data set for the copperhead snake (Agkistrodon contortrix) and address the population genomic impacts of recent and widespread landscape modification across an ~1,000‐km² region of eastern Kentucky, USA. Nonspatial population‐based assignment and clustering methods supported little to no population structure. However, using individual‐based spatial autocorrelation approaches we found evidence for genetic structuring which closely follows the path of a historically important highway which experienced high traffic volumes from c. 1920 to 1970 before losing most traffic to a newly constructed alternative route. We found no similar spatial genomic signatures associated with more recently constructed highways or surface mining activity, although a time lag effect may be responsible for the lack of any emergent spatial genetic patterns. Subsampling of our SNP data set suggested that similar results could be obtained with as few as 250 SNPs, and a range of thresholds for missing data exhibited limited impacts on the spatial patterns we detected. While we were not able to estimate relative effects of land uses or precise time lags, our findings highlight the importance of temporal factors in landscape genetics approaches, and suggest the potential advantages of genomic data sets and fine‐scale, spatially informed approaches for quantifying subtle genetic patterns in temporally complex landscapes.     

Heavy metal contamination and health risks of indoor dust around Xinqiao Mining Area,Tongling, China

Abstract

In this study, the concentrations and health risks of heavy metals (Cu, Pb, Zn, Ni, Co, Cd, and Cr) in indoor dust are investigated in the vicinity of the Xinqiao mining area, Tongling, China. Results indicate that heavy metals except Co were clearly enriched in indoor dust. Especially Cd was extremely enriched, followed by Zn, Cu, and Pb. However, no significant regional differences (p?>?0.05) were found in other elemental contents aside from Cu. Statistical analysis revealed that metal elements except Co were presumed to originate primarily from mining activities. Health risk assessment indicated that the hazard quotients and hazard indices of all studied metal elements were less than 1 and thus posed no potential noncancer health risks to adults and children. Moreover, the cancer risks of Ni, Cr, Cd, and Co were within acceptable ranges, implying no cancer risk to local residents; however, the noncarcinogenic risk of Pb and the carcinogenic risk of Cr and Cd warrant close attention.     

A text mining analysis of the climate change literature in industrial ecology

The literature on climate change research has evolved tremendously since the 1990s. The goal of this study is to use text mining to review the climate change literature and study the evolution of the main trends over time. Specific keywords from articles published in the special issue “ Industrial Ecology for Climate Change Adaptation and Resilience” in the Journal of Industrial Ecology are first selected. Details of over 35,000 publications containing these keywords are downloaded from the Web of Science from 1990 to 2018. The number of publications and co‐occurrence of keywords are analyzed. Moreover, latent Dirichlet allocation (LDA)—a probabilistic approach that can retrieve topics from large and unstructured text documents—is applied on the abstracts to uncover the main topics (consisting of new terms) that naturally emerge from them. The evolution in time of the importance of some emerging topics is then analyzed on the basis of their relative frequency. Overall, a rapid growth in climate change publications is observed. Terms such as “climate change adaptation” appear on the rise, whereas other terms are declining such as “pollution.” Moreover, several terms tend to co‐occur frequently, such as “climate change adaptation” and “resilience.” The database collected and the LiTCoF (Literature Topic Co‐occurrence and Frequency) Python‐based tool developed for this study are also made openly accessible. This article met the requirements for a gold – gold JIE data openness badge described http://jie.click/badges .     

Machine learning based modeling of households: A regionalized bottom‐up approach to investigate consumption‐induced environmental impacts

As major drivers of economy, households induce a large share of worldwide environmental impacts. The variability of local consumption patterns and associated environmental impacts needs to be quantified as an important starting point to devise targeted measures aimed at reducing household environmental footprints. The goal of this article is the development and appraisal of a comprehensive regionalized bottom‐up model that assesses realistic environmental profiles for individual households in a specific region. For this purpose, a physically based building energy model, the results of an agent‐based transport simulation, and a data‐driven household consumption model were interlinked within a new probability‐based classification framework and applied to the case of Switzerland. The resulting model predicts the demands in about 400 different consumption areas for each Swiss household by considering its particular circumstances and produces a realistic picture of variability in household environmental footprints. An analysis of the model results on a municipal level reveals per‐capita income, population density, buildings' age, and household structure as possible drivers of municipal carbon footprints. While higher‐emission municipalities are located in rural areas and tend to show higher shares of older buildings, lower‐emission communities have larger proportions of families and can be found in highly populated regions by trend. However, the opposing effects of various variables observed in this analysis confirm the importance of a model that is able to capture regional distinctions. The overall model constitutes a comprehensive information base supporting policymakers in understanding consumption patterns in their region and deriving environmental strategies tailored to their specific population.