An explainable and efficient deep learning framework for video anomaly detection

Cluster Computing - Deep learning-based video anomaly detection methods have drawn significant attention in the past few years due to their superior performance. However, almost all the leading...     

Emerging deep learning methods for single-cell RNA-seq data analysis

Deep learning is making major breakthrough in several areas of bioinformatics. Anticipating that this will occur soon for the single-cell RNA-seq data analysis, we review newly published deep learning methods that help tackle computational challenges. Autoencoders are found to be the dominant approach. However, methods based on deep generative models such as generative adversarial networks (GANs) are also emerging in this area.     

Student-centered introductory biology course: evidence for deep learning

Our interpretative study that was carried out in a science and engineering oriented university examined the ways students in an introductory biology course perceived their learning in the course that was substantially changed to allow student-centered learning. The instructional change was framed by the view of learning as a sociocultural activity as well as a cognitive process that can take place face-to-face or through online interaction. Most of the lectures were substituted with individual learning and project-based, small-group learning that lasted one month. Data were collected through interviews with students and instructors, and through observations. In the paper, we show evidence for deep learning that was associated by the students and the instructors with short-term, meaningful activities in a setting that included collaborative peer learning; and replacing most lectures by small group learning that ended in a mini-conference. Deep learning was evidenced by the ways students reflected on how they organised and applied knowledge using deep learning strategies.     

Toxic gas dispersion prediction for point source emission using deep learning method

Abstract

Accurate and rapid toxic gas concentration prediction model plays an important role in emergency aid of sudden gas leak. However, it is difficult for existing dispersion model to achieve accuracy and efficiency requirements at the same time. Although some researchers have considered developing new forecasting models with traditional machine learning, such as back propagation (BP) neural network, support vector machine (SVM), the prediction results obtained from such models need to be improved still in terms of accuracy. Then new prediction models based on deep learning are proposed in this paper. Deep learning has obvious advantages over traditional machine learning in prediction and classification. Deep belief networks (DBNs) as well as convolution neural networks (CNNs) are used to build new dispersion models here. Both models are compared with Gaussian plume model, computation fluid dynamics (CFD) model and models based on traditional machine learning in terms of accuracy, prediction time, and computation time. The experimental results turn out that CNNs model performs better considering all evaluation indexes.     

基于深度学习的蛋白质建模与设计

随着蛋白质序列及结构数据的大量累积,在获得了大量描述性信息之后如何有效利用海量数据,从已有数据中高效提取信息并且应用到下游任务当中就成为了研究者亟待解决的问题。蛋白质的设计可使新蛋白的研发不再受限于实验条件,这对药物靶点预测、新药研发和材料设计等领域具有重要意义。深度学习作为一种高效的数据特征提取方法,可以通过它对蛋白质数据进行建模,进而加入先验信息对蛋白质进行设计。故此基于深度学习的蛋白质设计就成为一个具有广阔前景的研究领域。文中主要阐述基于深度学习的蛋白质序列与结构数据的建模和设计方法。详述该方法的策略、原理、适用范围、应用实例。讨论了深度学习方法在本领域的应用前景及局限性,以期为相关研究提供参考。     

Wildlife surveillance using deep learning methods

相似文献   

AggMapNet: enhanced and explainable low-sample omics deep learning with feature-aggregated multi-channel networks

Omics-based biomedical learning frequently relies on data of high-dimensions (up to thousands) and low-sample sizes (dozens to hundreds), which challenges efficient deep learning (DL) algorithms, particularly for low-sample omics investigations. Here, an unsupervised novel feature aggregation tool AggMap was developed to Aggregate and Map omics features into multi-channel 2D spatial-correlated image-like feature maps (Fmaps) based on their intrinsic correlations. AggMap exhibits strong feature reconstruction capabilities on a randomized benchmark dataset, outperforming existing methods. With AggMap multi-channel Fmaps as inputs, newly-developed multi-channel DL AggMapNet models outperformed the state-of-the-art machine learning models on 18 low-sample omics benchmark tasks. AggMapNet exhibited better robustness in learning noisy data and disease classification. The AggMapNet explainable module Simply-explainer identified key metabolites and proteins for COVID-19 detections and severity predictions. The unsupervised AggMap algorithm of good feature restructuring abilities combined with supervised explainable AggMapNet architecture establish a pipeline for enhanced learning and interpretability of low-sample omics data.     

Transcriptional landscape of DNA repair genes underpins a pan-cancer prognostic signature associated with cell cycle dysregulation and tumor hypoxia

Overactive DNA repair contributes to therapeutic resistance in cancer. However, pan-cancer comparative studies investigating the contribution of all DNA repair genes in cancer progression employing an integrated approach have remained limited. We performed a multi-cohort retrospective analysis to determine the prognostic significance of 138 DNA repair genes in 16 cancer types (n = 16,225). Cox proportional hazards analyses revealed a significant variation in the number of prognostic genes between cancers; 81 genes were prognostic in clear cell renal cell carcinoma while only two genes were prognostic in glioblastoma. We reasoned that genes that were commonly prognostic in highly correlated cancers revealed by Spearman’s correlation analysis could be harnessed as a molecular signature for risk assessment. A 10-gene signature, uniting prognostic genes that were common in highly correlated cancers, was significantly associated with overall survival in patients with clear cell renal cell (P < 0.0001), papillary renal cell (P = 0.0007), liver (P = 0.002), lung (P = 0.028), pancreas (P = 0.00013) or endometrial (P = 0.00063) cancers. Receiver operating characteristic analyses revealed that a combined model of the 10-gene signature and tumor staging outperformed either classifier when considered alone. Multivariate Cox regression models incorporating additional clinicopathological features showed that the signature was an independent predictor of overall survival. Tumor hypoxia is associated with adverse outcomes. Consistent across all six cancers, patients with high 10-gene and high hypoxia scores had significantly higher mortality rates compared to those with low 10-gene and low hypoxia scores. Functional enrichment analyses revealed that high mortality rates in patients with high 10-gene scores were attributable to an overproliferation phenotype. Death risk in these patients was further exacerbated by concurrent mutations of a cell cycle checkpoint protein, TP53. The 10-gene signature identified tumors with heightened DNA repair ability. This information has the potential to radically change prognosis through the use of adjuvant DNA repair inhibitors with chemotherapeutic drugs.     

LV-GAN: A deep learning approach for limited-view optoacoustic imaging based on hybrid datasets

The optoacoustic imaging (OAI) methods are rapidly evolving for resolving optical contrast in medical imaging applications. In practice, measurement strategies are commonly implemented under limited-view conditions due to oversized image objectives or system design limitations. Data acquired by limited-view detection may impart artifacts and distortions in reconstructed optoacoustic (OA) images. We propose a hybrid data-driven deep learning approach based on generative adversarial network (GAN), termed as LV-GAN, to efficiently recover high quality images from limited-view OA images. Trained on both simulation and experiment data, LV-GAN is found capable of achieving high recovery accuracy even under limited detection angles less than 60^°. The feasibility of LV-GAN for artifact removal in biological applications was validated by ex vivo experiments based on two different OAI systems, suggesting high potential of a ubiquitous use of LV-GAN to optimize image quality or system design for different scanners and application scenarios.     

A deep learning model for predicting optimal distance range in crosslinking mass spectrometry data

Macromolecular assemblies play an important role in all cellular processes. While there has recently been significant progress in protein structure prediction based on deep learning, large protein complexes cannot be predicted with these approaches. The integrative structure modeling approach characterizes multi-subunit complexes by computational integration of data from fast and accessible experimental techniques. Crosslinking mass spectrometry is one such technique that provides spatial information about the proximity of crosslinked residues. One of the challenges in interpreting crosslinking datasets is designing a scoring function that, given a structure, can quantify how well it fits the data. Most approaches set an upper bound on the distance between Cα atoms of crosslinked residues and calculate a fraction of satisfied crosslinks. However, the distance spanned by the crosslinker greatly depends on the neighborhood of the crosslinked residues. Here, we design a deep learning model for predicting the optimal distance range for a crosslinked residue pair based on the structures of their neighborhoods. We find that our model can predict the distance range with the area under the receiver-operator curve of 0.86 and 0.7 for intra- and inter-protein crosslinks, respectively. Our deep scoring function can be used in a range of structure modeling applications.     

A new deep learning method for image deblurring in optical microscopic systems

Deconvolution is the most commonly used image processing method in optical imaging systems to remove the blur caused by the point‐spread function (PSF). While this method has been successful in deblurring, it suffers from several disadvantages, such as slow processing time due to multiple iterations required to deblur and suboptimal in cases where the experimental operator chosen to represent PSF is not optimal. In this paper, we present a deep‐learning‐based deblurring method that is fast and applicable to optical microscopic imaging systems. We tested the robustness of proposed deblurring method on the publicly available data, simulated data and experimental data (including 2D optical microscopic data and 3D photoacoustic microscopic data), which all showed much improved deblurred results compared to deconvolution. We compared our results against several existing deconvolution methods. Our results are better than conventional techniques and do not require multiple iterations or pre‐determined experimental operator. Our method has several advantages including simple operation, short time to compute, good deblur results and wide application in all types of optical microscopic imaging systems. The deep learning approach opens up a new path for deblurring and can be applied in various biomedical imaging fields.     

DCAMCP: A deep learning model based on capsule network and attention mechanism for molecular carcinogenicity prediction

The carcinogenicity of drugs can have a serious impact on human health, so carcinogenicity testing of new compounds is very necessary before being put on the market. Currently, many methods have been used to predict the carcinogenicity of compounds. However, most methods have limited predictive power and there is still much room for improvement. In this study, we construct a deep learning model based on capsule network and attention mechanism named DCAMCP to discriminate between carcinogenic and non-carcinogenic compounds. We train the DCAMCP on a dataset containing 1564 different compounds through their molecular fingerprints and molecular graph features. The trained model is validated by fivefold cross-validation and external validation. DCAMCP achieves an average accuracy (ACC) of 0.718 ± 0.009, sensitivity (SE) of 0.721 ± 0.006, specificity (SP) of 0.715 ± 0.014 and area under the receiver-operating characteristic curve (AUC) of 0.793 ± 0.012. Meanwhile, comparable results can be achieved on an external validation dataset containing 100 compounds, with an ACC of 0.750, SE of 0.778, SP of 0.727 and AUC of 0.811, which demonstrate the reliability of DCAMCP. The results indicate that our model has made progress in cancer risk assessment and could be used as an efficient tool in drug design.     

Microscopic image recognition of diatoms based on deep learning

Diatoms are a crucial component in the study of aquatic ecosystems and ancient environmental records. However, traditional methods for identifying diatoms, such as morphological taxonomy and molecular detection, are costly, are time consuming, and have limitations. To address these issues, we developed an extensive collection of diatom images, consisting of 7983 images from 160 genera and 1042 species, which we expanded to 49,843 through preprocessing, segmentation, and data augmentation. Our study compared the performance of different algorithms, including backbones, batch sizes, dynamic data augmentation, and static data augmentation on experimental results. We determined that the ResNet152 network outperformed other networks, producing the most accurate results with top-1 and top-5 accuracies of 85.97% and 95.26%, respectively, in identifying 1042 diatom species. Additionally, we propose a method that combines model prediction and cosine similarity to enhance the model's performance in low-probability predictions, achieving an 86.07% accuracy rate in diatom identification. Our research contributes significantly to the recognition and classification of diatom images and has potential applications in water quality assessment, ecological monitoring, and detecting changes in aquatic biodiversity.     

Breast glandularity and mean glandular dose assessment using a deep learning framework: Virtual patients study

PurposeBreast dosimetry in mammography is an important aspect of radioprotection since women are exposed periodically to ionizing radiation due to breast cancer screening programs. Mean glandular dose (MGD) is the standard quantity employed for the establishment of dose reference levels in retrospective population studies. However, MGD calculations requires breast glandularity estimation. This work proposes a deep learning framework for volume glandular fraction (VGF) estimations based on mammography images, which in turn are converted to glandularity values for MGD calculations.Methods208 virtual breast phantoms were generated and compressed computationally. The mammography images were obtained with Monte Carlo simulations (MC-GPU code) and a ray-tracing algorithm was employed for labeling the training data. The architectures of the neural networks are based on the XNet and multilayer perceptron, adapted for each task. The network predictions were compared with the ground truth using the coefficient of determination (r²).ResultsThe results have shown a good agreement for inner breast segmentation (r² = 0.999), breast volume prediction (r² = 0.982) and VGF prediction (r² = 0.935). Moreover, the DgN coefficients using the predicted VGF for the virtual population differ on average 1.3% from the ground truth values. Afterwards with the obtained DgN coefficients, the MGD values were estimated from exposure factors extracted from the DICOM header of a clinical cohort, with median(75 percentile) values of 1.91(2.45) mGy.ConclusionWe successfully implemented a deep learning framework for VGF and MGD calculations for virtual breast phantoms.     

DeepPhagy: a deep learning framework for quantitatively measuring autophagy activity in Saccharomyces cerevisiae

ABSTRACT

Seeing is believing. The direct observation of GFP-Atg8 vacuolar delivery under confocal microscopy is one of the most useful end-point measurements for monitoring yeast macroautophagy/autophagy. However, manually labelling individual cells from large-scale sets of images is time-consuming and labor-intensive, which has greatly hampered its extensive use in functional screens. Herein, we conducted a time-course analysis of nitrogen starvation-induced autophagy in wild-type and knockout mutants of 35 AuTophaGy-related (ATG) genes in Saccharomyces cerevisiae and obtained 1,944 confocal images containing > 200,000 cells. We manually labelled 8,078 autophagic and 18,493 non-autophagic cells as a benchmark dataset and developed a new deep learning tool for autophagy (DeepPhagy), which exhibited superior accuracy in recognizing autophagic cells compared to other existing methods, with an area under the curve (AUC) value of 0.9710 from 10-fold cross-validations. We further used DeepPhagy to automatically analyze all the images and quantitatively classified the autophagic phenotypes of the 35 atg knockout mutants into 3 classes. The high consistency in our computational and biochemical results indicated the reliability of DeepPhagy for measuring autophagic activity. Moreover, we used DeepPhagy to analyze 3 additional types of autophagic phenotypes, including the targeting of Atg1-GFP to the vacuole, the vacuolar delivery of GFP-Atg19, and the disintegration of autophagic bodies indicated by GFP-Atg8, all with satisfying accuracies. Taken together, our study not only enables the GFP-Atg8 ?uorescence assay to become a quantitative measurement for analyzing autophagic phenotypes in S. cerevisiae but also demonstrates that deep learning-based methods could potentially be applied to different types of autophagy.

Abbreviations: Ac: accuracy; ALP: alkaline phosphatase; ALR: autophagic lysosomal reformation; ATG: AuTophaGy-related; AUC: area under the curve; CNN: convolutional neural network; Cvt: cytoplasm-to-vacuole targeting; DeepPhagy: deep learning for autophagy; fc_2: second fully connected; GFP: green fluorescent protein; MAP1LC3/LC3: microtubule-associated protein 1 light chain 3 beta; HAT: histone acetyltransferase; HemI: Heat map Illustrator; JRE: Java Runtime Environment; KO: knockout; LRN: local response normalization; MCC: Mathew Correlation Coefficient; OS: operating system; PAS: phagophore assembly site; PC: principal component; PCA: principal component analysis; PPI: protein-protein interaction; Pr: precision; QPSO: Quantum-behaved Particle Swarm Optimization; ReLU: rectified linear unit; RF: random forest; ROC: receiver operating characteristic; ROI: region of interest; SD: systematic derivation; SGD: stochastic gradient descent; Sn: sensitivity; Sp: specificity; SRG: seeded region growing; t-SNE: t-distributed stochastic neighbor embedding; 2D: 2-dimensional; WT: wild-type.