Automatic and accurate estimation of disease severity is essential for food security, disease management, and yield loss prediction. Deep learning, the latest breakthrough in computer vision, is promising for fine-grained disease severity classification, as the method avoids the labor-intensive feature engineering and threshold-based segmentation. Using the apple black rot images in the PlantVillage dataset, which are further annotated by botanists with four severity stages as ground truth, a series of deep convolutional neural networks are trained to diagnose the severity of the disease. The performances of shallow networks trained from scratch and deep models fine-tuned by transfer learning are evaluated systemically in this paper. The best model is the deep VGG16 model trained with transfer learning, which yields an overall accuracy of 90.4% on the hold-out test set. The proposed deep learning model may have great potential in disease control for modern agriculture.
Online reviews play a crucial role in today's electronic commerce. It is desirable for a customer to read reviews of products or stores before making the decision of what or from where to buy. Due to the pervasive spam reviews, customers can be misled to buy low-quality products, while decent stores can be defamed by malicious reviews. We observe that, in reality, a great portion (> 90% in the data we study) of the reviewers write only one review (singleton review). These reviews are so enormous in number that they can almost determine a store's rating and impression. However, existing methods did not examine this larger part of the reviews. Are most of these singleton reviews truthful ones? If not, how to detect spam reviews in singleton reviews? We call this problem singleton review spam detection.To address this problem, we observe that the normal reviewers' arrival pattern is stable and uncorrelated to their rating pattern temporally. In contrast, spam attacks are usually bursty and either positively or negatively correlated to the rating. Thus, we propose to detect such attacks via unusually correlated temporal patterns. We identify and construct multidimensional time series based on aggregate statistics, in order to depict and mine such correlations. In this way, the singleton review spam detection problem is mapped to a abnormally correlated pattern detection problem. We propose a hierarchical algorithm to robustly detect the time windows where such attacks are likely to have happened. The algorithm also pinpoints such windows in different time resolutions to facilitate faster human inspection. Experimental results show that the proposed method is effective in detecting singleton review attacks. We discover that singleton review is a significant source of spam reviews and largely affects the ratings of online stores.
We reported HIVID (high-throughput Viral Integration Detection), a novel experimental and computational method to detect the location of Hepatitis B Virus (HBV) integration breakpoints in Hepatocellular Carcinoma (HCC) genome. In this method, the fragments with HBV sequence were enriched by a set of HBV probes and then processed to high-throughput sequencing. In order to evaluate the performance of HIVID, we compared the results of HIVID with that of whole genome sequencing method (WGS) in 28 HCC tumors. We detected a total of 246 HBV integration breakpoints in HCC genome, 113 out of which were within 400bp upstream or downstream of 125 breakpoints identified by WGS method, covering 89.3% (125/140) of total breakpoints. The integration was located in the gene TERT, MLL4, and CCNE1. In addition, we discovered 133 novel breakpoints missed by WGS method, with 66.7% (10/15) of validation rate. Our study shows HIVID is a cost-effective methodology with high specificity and sensitivity to identify viral integration in human genome.
Single-cell genomic analysis has grown rapidly in recent years and finds widespread applications in various fields of biology, including cancer biology, development, immunology, pre-implantation genetic diagnosis, and neurobiology. To date, the amplification bias, amplification uniformity and reproducibility of the three major single cell whole genome amplification methods (GenomePlex WGA4, MDA and MALBAC) have not been systematically investigated using mammalian cells. In this study, we amplified genomic DNA from individual hippocampal neurons using three single-cell DNA amplification methods, and sequenced them at shallow depth. We then systematically evaluated the GC-bias, reproducibility, and copy number variations among individual neurons. Our results showed that single-cell genome sequencing results obtained from the MALBAC and WGA4 methods are highly reproducible and have a high success rate. The MALBAC displays significant biases towards high GC content. We then attempted to correct the GC bias issue by developing a bioinformatics pipeline, which allows us to call CNVs in single cell sequencing data, and chromosome level and sub-chromosomal level CNVs among individual neurons can be detected. We also proposed a metric to determine the CNV detection limits. Overall, MALBAC and WGA4 have better performance than MDA in detecting CNVs.
Azo dyes are recalcitrant and refractory pollutants that constitute a significant menace to the environment. The present study is focused on exploring the capability of Bacillus sp. strain UN2 for application in methyl red (MR) degradation. Effects of physicochemical parameters (pH of medium, temperature, initial concentration of dye, and composition of the medium) were studied in detail. The suitable pH and temperature range for MR degradation by strain UN2 were respectively 7.0-9.0 and 30-40 °C, and the optimal pH value and temperature were respectively 8.0 and 35 °C. Mg(2+) and Mn(2+) (1 mM) were found to significantly accelerate the MR removal rate, while the enhancement by either Fe(3+) or Fe(2+) was slight. Under the optimal degradation conditions, strain UN2 exhibited greater than 98 % degradation of the toxic azo dye MR (100 ppm) within 30 min. Analysis of samples from decolorized culture flasks confirmed biodegradation of MR into two prime metabolites: N,N'dimethyl-p-phenyle-nediamine and 2-aminobenzoic acid. A study of the enzymes responsible for the biodegradation of MR, in the control and cells obtained during (10 min) and after (30 min) degradation, showed a significant increase in the activities of azoreductase, laccase, and NADH-DCIP reductase. Furthermore, a phytotoxicity analysis demonstrated that the germination inhibition was almost eliminated for both the plants Triticum aestivum and Sorghum bicolor by MR metabolites at 100 mg/L concentration, yet the germination inhibition of parent dye was significant. Consequently, the high efficiency of MR degradation enables this strain to be a potential candidate for bioremediation of wastewater containing MR.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.