RESUMO
This manuscript describes the development of a resources module that is part of a learning platform named 'NIGMS Sandbox for Cloud-based Learning' https://github.com/NIGMS/NIGMS-Sandbox. The overall genesis of the Sandbox is described in the editorial NIGMS Sandbox at the beginning of this Supplement. This module delivers learning materials on implementing deep learning algorithms for biomedical image data in an interactive format that uses appropriate cloud resources for data access and analyses. Biomedical-related datasets are widely used in both research and clinical settings, but the ability for professionally trained clinicians and researchers to interpret datasets becomes difficult as the size and breadth of these datasets increases. Artificial intelligence, and specifically deep learning neural networks, have recently become an important tool in novel biomedical research. However, use is limited due to their computational requirements and confusion regarding different neural network architectures. The goal of this learning module is to introduce types of deep learning neural networks and cover practices that are commonly used in biomedical research. This module is subdivided into four submodules that cover classification, augmentation, segmentation and regression. Each complementary submodule was written on the Google Cloud Platform and contains detailed code and explanations, as well as quizzes and challenges to facilitate user training. Overall, the goal of this learning module is to enable users to identify and integrate the correct type of neural network with their data while highlighting the ease-of-use of cloud computing for implementing neural networks. This manuscript describes the development of a resource module that is part of a learning platform named ``NIGMS Sandbox for Cloud-based Learning'' https://github.com/NIGMS/NIGMS-Sandbox. The overall genesis of the Sandbox is described in the editorial NIGMS Sandbox [1] at the beginning of this Supplement. This module delivers learning materials on the analysis of bulk and single-cell ATAC-seq data in an interactive format that uses appropriate cloud resources for data access and analyses.
Assuntos
Aprendizado Profundo , Redes Neurais de Computação , Humanos , Pesquisa Biomédica , Algoritmos , Computação em NuvemRESUMO
BACKGROUND: Twitter provides a valuable platform for the surveillance and monitoring of public health topics; however, manually categorizing large quantities of Twitter data is labor intensive and presents barriers to identify major trends and sentiments. Additionally, while machine and deep learning approaches have been proposed with high accuracy, they require large, annotated data sets. Public pretrained deep learning classification models, such as BERTweet, produce higher-quality models while using smaller annotated training sets. OBJECTIVE: This study aims to derive and evaluate a pretrained deep learning model based on BERTweet that can identify tweets relevant to vaping, tweets (related to vaping) of commercial nature, and tweets with provape sentiment. Additionally, the performance of the BERTweet classifier will be compared against a long short-term memory (LSTM) model to show the improvements a pretrained model has over traditional deep learning approaches. METHODS: Twitter data were collected from August to October 2019 using vaping-related search terms. From this set, a random subsample of 2401 English tweets was manually annotated for relevance (vaping related or not), commercial nature (commercial or not), and sentiment (positive, negative, or neutral). Using the annotated data, 3 separate classifiers were built using BERTweet with the default parameters defined by the Simple Transformer application programming interface (API). Each model was trained for 20 iterations and evaluated with a random split of the annotated tweets, reserving 10% (n=165) of tweets for evaluations. RESULTS: The relevance, commercial, and sentiment classifiers achieved an area under the receiver operating characteristic curve (AUROC) of 94.5%, 99.3%, and 81.7%, respectively. Additionally, the weighted F1 scores of each were 97.6%, 99.0%, and 86.1%, respectively. We found that BERTweet outperformed the LSTM model in the classification of all categories. CONCLUSIONS: Large, open-source deep learning classifiers, such as BERTweet, can provide researchers the ability to reliably determine if tweets are relevant to vaping; include commercial content; and include positive, negative, or neutral content about vaping with a higher accuracy than traditional natural language processing deep learning models. Such enhancement to the utilization of Twitter data can allow for faster exploration and dissemination of time-sensitive data than traditional methodologies (eg, surveys, polling research).
RESUMO
Mitochondria are dynamic organelles that differ significantly in their morphologies across cell types, reflecting specific cellular needs and stages in development. Despite the wide biological significance in disease and in health, delineating mitochondrial morphologies in complex systems remains challenging. Here, we present the Mitochondrial Cellular Phenotype (MitoCellPhe) tool developed for quantifying mitochondrial morphologies and demonstrate its utility in delineating differences in mitochondrial morphologies in a human fibroblast and human induced pluripotent stem cell (hiPSC) line. MitoCellPhe generates 24 parameters, allowing for a comprehensive analysis of mitochondrial structures and importantly allows for quantification to be performed on mitochondria in images containing single cells or clusters of cells. With this tool, we were able to validate previous findings that show networks of mitochondria in healthy fibroblast cell lines and a more fragmented morphology in hiPSCs. Using images generated from control and diseased fibroblasts and hiPSCs, we also demonstrate the efficacy of the toolset in delineating differences in morphologies between healthy and the diseased state in both stem cell (hiPSC) and differentiated fibroblast cells. Our results demonstrate that MitoCellPhe enables high-throughput, sensitive, detailed, and quantitative mitochondrial morphological assessment and thus enables better biological insights into mitochondrial dynamics in health and disease.
Assuntos
Fibroblastos/patologia , Processamento de Imagem Assistida por Computador , Células-Tronco Pluripotentes Induzidas/patologia , Microscopia de Fluorescência , Mitocôndrias/patologia , Dinâmica Mitocondrial , Forma das Organelas , Design de Software , Linhagem Celular , Ensaios de Triagem em Larga Escala , Humanos , FenótipoRESUMO
Deep learning models have been shown to be effective for material analysis, a subfield of computer vision, on natural images. In medicine, deep learning systems have been shown to more accurately analyze radiography images than algorithmic approaches and even experts. However, one major roadblock to applying deep learning-based material analysis on radiography images is a lack of material annotations accompanying image sets. To solve this, we first introduce an automated procedure to augment annotated radiography images into a set of material samples. Next, using a novel Siamese neural network that compares material sample pairs, called D-CNN, we demonstrate how to learn a perceptual distance metric between material categories. This system replicates the actions of human annotators by discovering attributes that encode traits that distinguish materials in radiography images. Finally, we update and apply MAC-CNN, a material recognition neural network, to demonstrate this system on a dataset of knee X-rays and brain MRIs with tumors. Experiments show that this system has strong predictive power on these radiography images, achieving 92.8% accuracy at predicting the material present in a local region of an image. Our system also draws interesting parallels between human perception of natural materials and materials in radiography images.
RESUMO
In this article, we propose a new approach to analyze large genomics data. We considered individual genetic variants as pixels in an image and transformed a collection of variants into an artificial image object (AIO), which could be classified as a regular image by CNN algorithms. Using schizophrenia as a case study, we demonstrate the principles and their applications with 3 datasets. With 4,096 SNVs, the CNN models achieved an accuracy of 0.678 ± 0.007 and an AUC of 0.738 ± 0.008 for the diagnosis phenotype. With 44,100 SNVs, the models achieved class-specific accuracies of 0.806 ± 0.032 and 0.820 ± 0.049, and AUCs of 0.930 ± 0.017 and 0.867 ± 0.040 for the bottom and top classes stratified by the patient's polygenic risk scores. These results suggest that, once transformed to images, large genomics data can be analyzed effectively with image classification algorithms.
RESUMO
Signaling pathways describe a group of molecules in a cell that collaborate to control one or more cell functions, such as cell division or cell death. The pathways communicate by sending signals between molecules, and this process is repeated until the terminal molecule is activated and the cell function is executed. Signaling pathways are often represented as directed graphs, which does not provide enough information when modeling cell functions and reactions. Recently, directed hypergraphs have been proposed to more accurately represent reactions such as protein activation and interaction. To further improve the representation of signaling pathways, time dependency must be considered to improve the representation of cell signaling at any given time. In this paper, the importance of time dependency in modeling signaling pathways is presented. An algorithm that finds the shortest a priori path using time-dependent hypergraphs to more robustly model signaling pathways is adopted. The shortest time-dependent hyperpaths representing signaling pathways are an improvement to the recent adoption of hypergraphs representing these pathways. The results display the improved representation of signaling pathways and motivate the adoption of time-dependent signaling hypergraphs.
Assuntos
Modelos Biológicos , Transdução de Sinais/fisiologia , Biologia de Sistemas/métodos , Fatores de Tempo , AlgoritmosRESUMO
High utility itemset mining has become an important and critical operation in the Data Mining field. High utility itemset mining generates more profitable itemsets and the association among these itemsets, to make business decisions and strategies. Although, high utility is important, it is not the sole measure to decide efficient business strategies such as discount offers. It is very important to consider the pattern of itemsets based on the frequency as well as utility to predict more profitable itemsets. For example, in a supermarket or restaurant, beverages like champagne or wine might generate high utility (profit), but also sell less frequently compared to other beverages like soda or beer. In previous studies, it is observed that people who buy milk, bread, or diapers from a supermarket, also tend to buy beer or soda. But the items like milk, diapers, beer, or soda generate less utility (profit value) compared to beverages like champagne or wine. If we combine items like champagne or wine having high utility but less frequency, with the frequently sold items like milk, diaper, or beer, we can increase the utility of the transaction by providing some discount offers on champagne or wine. In this paper, we are integrating low-frequency itemsets with high-frequency itemsets, both having low or high utility, and provide different association rules for this combination of itemsets. In this way, we can generate a more accurate measure of pattern mining for various business strategies.