Deep Learning-Based construction of a Drug-Like compound database and its application in virtual screening of HsDHODH inhibitors

The process of virtual screening relies heavily on the databases, but it is disadvantageous to conduct virtual screening based on commercial databases with patent-protected compounds, high compound toxicity and side effects. Therefore, this paper utilizes generative recurrent neural networks (RNN) containing long short-term memory (LSTM) cells to learn the properties of drug compounds in the DrugBank, aiming to obtain a new and virtual screening compounds database with drug-like properties. Ultimately, a compounds database consisting of 26,316 compounds is obtained by this method. To evaluate the potential of this compounds database, a series of tests are performed, including chemical space, ADME properties, compound fragmentation, and synthesizability analysis. As a result, it is proved that the database is equipped with good drug-like properties and a relatively new backbone, its potential in virtual screening is further tested. Finally, a series of seedling compounds with completely new backbones are obtained through docking and binding free energy calculations.

Exploring weight initialization, diversity of solutions, and degradation in recurrent neural networks trained for temporal and decision-making tasks

Recurrent Neural Networks (RNNs) are frequently used to model aspects of brain function and structure. In this work, we trained small fully-connected RNNs to perform temporal and flow control tasks with time-varying stimuli. Our results show that different RNNs can solve the same task by converging to different underlying dynamics and also how the performance gracefully degrades as either network size is decreased, interval duration is increased, or connectivity damage is induced. For the considered tasks, we explored how robust the network obtained after training can be according to task parameterization. In the process, we developed a framework that can be useful to parameterize other tasks of interest in computational neuroscience. Our results are useful to quantify different aspects of the models, which are normally used as black boxes and need to be understood in order to model the biological response of cerebral cortex areas.

SGORNN: Combining scalar gates and orthogonal constraints in recurrent networks

Recurrent Neural Network (RNN) models have been applied in different domains, producing high accuracies on time-dependent data. However, RNNs have long suffered from exploding gradients during training, mainly due to their recurrent process. In this context, we propose a variant of the scalar gated FastRNN architecture, called Scalar Gated Orthogonal Recurrent Neural Networks (SGORNN). SGORNN utilizes orthogonal matrices at the recurrent step. Our experiments evaluate SGORNN using two recently proposed orthogonal parametrizations for the recurrent weights of an RNN. We present a constraint on the scalar gates of SGORNN, which is easily enforced at training time to provide a probabilistic generalization gap which grows linearly with the length of sequences processed. Next, we provide bounds on the gradients of SGORNN to show the impossibility of exponentially exploding gradients through time. Our experimental results on the addition problem confirm that our combination of orthogonal and scalar gated RNNs are able to outperform other orthogonal RNNs and LSTM on long sequences. We further evaluate SGORNN on the HAR-2 classification task, where it improves upon the accuracy of several models using far fewer parameters than standard RNNs. Finally, we evaluate SGORNN on the Penn Treebank word-level language modeling task, where it again outperforms its related architectures and shows comparable performance to LSTM using far less parameters. Overall, SGORNN shows higher representation capacity than the other orthogonal RNNs tested, suffers from less overfitting than other models in our experiments, benefits from a decrease in parameter count, and alleviates exploding gradients during backpropagation through time.

Using Recurrent Neural Networks for Predicting Type-2 Diabetes from Genomic and Tabular Data

The development of genomic technology for smart diagnosis and therapies for various diseases has lately been the most demanding area for computer-aided diagnostic and treatment research. Exponential breakthroughs in artificial intelligence and machine intelligence technologies could pave the way for identifying challenges afflicting the healthcare industry. Genomics is paving the way for predicting future illnesses, including cancer, Alzheimer's disease, and diabetes. Machine learning advancements have expedited the pace of biomedical informatics research and inspired new branches of computational biology. Furthermore, knowing gene relationships has resulted in developing more accurate models that can effectively detect patterns in vast volumes of data, making classification models important in various domains. Recurrent Neural Network models have a memory that allows them to quickly remember knowledge from previous cycles and process genetic data. The present work focuses on type 2 diabetes prediction using gene sequences derived from genomic DNA fragments through automated feature selection and feature extraction procedures for matching gene patterns with training data. The suggested model was tested using tabular data to predict type 2 diabetes based on several parameters. The performance of neural networks incorporating Recurrent Neural Network (RNN) components, Long Short-Term Memory (LSTM), and Gated Recurrent Units (GRU) was tested in this research. The model's efficiency is assessed using the evaluation metrics such as Sensitivity, Specificity, Accuracy, F1-Score, and Mathews Correlation Coefficient (MCC). The suggested technique predicted future illnesses with fair Accuracy. Furthermore, our research showed that the suggested model could be used in real-world scenarios and that input risk variables from an end-user Android application could be kept and evaluated on a secure remote server.

Self-supervised recurrent depth estimation with attention mechanisms

Depth estimation has been an essential task for many computer vision applications, especially in autonomous driving, where safety is paramount. Depth can be estimated not only with traditional supervised learning but also via a self-supervised approach that relies on camera motion and does not require ground truth depth maps. Recently, major improvements have been introduced to make self-supervised depth prediction more precise. However, most existing approaches still focus on single-frame depth estimation, even in the self-supervised setting. Since most methods can operate with frame sequences, we believe that the quality of current models can be significantly improved with the help of information about previous frames. In this work, we study different ways of integrating recurrent blocks and attention mechanisms into a common self-supervised depth estimation pipeline. We propose a set of modifications that utilize temporal information from previous frames and provide new neural network architectures for monocular depth estimation in a self-supervised manner. Our experiments on the KITTI dataset show that proposed modifications can be an effective tool for exploiting temporal information in a depth prediction pipeline.

Robust Brain State Decoding using Bidirectional Long Short Term Memory Networks in functional MRI

Decoding brain states of the underlying cognitive processes via learning discriminative feature representations has recently gained a lot of interest in brain imaging studies. Particularly, there has been an impetus to encode the dynamics of brain functioning by analyzing temporal information available in the fMRI data. Long-short term memory (LSTM), a class of machine learning model possessing a "memory" component, to retain previously seen temporal information, is increasingly being observed to perform well in various applications with dynamic temporal behavior, including brain state decoding. Because of the dynamics and inherent latency in fMRI BOLD responses, future temporal context is crucial. However, it is neither encoded nor captured by the conventional LSTM model. This paper performs robust brain state decoding via information encapsulation from both the past and future instances of fMRI data via bi-directional LSTM. This allows for explicitly modeling the dynamics of BOLD response without any delay adjustment. To this end, we utilize a bidirectional LSTM, wherein, the input sequence is fed in normal time-order for one LSTM network, and in the reverse time-order, for another. The two hidden activations of forward and reverse directions in bi-LSTM are collated to build the "memory" of the model and are used to robustly predict the brain states at every time instance. Working memory data from the Human Connectome Project (HCP) is utilized for validation and was observed to perform 18% better than it's unidirectional counterpart in terms of accuracy in predicting the brain states.

Generative Deep Learning for Targeted Compound Design

In the past few years, de novo molecular design has increasingly been using generative models from the emergent field of Deep Learning, proposing novel compounds that are likely to possess desired properties or activities. De novo molecular design finds applications in different fields ranging from drug discovery and materials sciences to biotechnology. A panoply of deep generative models, including architectures as Recurrent Neural Networks, Autoencoders, and Generative Adversarial Networks, can be trained on existing data sets and provide for the generation of novel compounds. Typically, the new compounds follow the same underlying statistical distributions of properties exhibited on the training data set Additionally, different optimization strategies, including transfer learning, Bayesian optimization, reinforcement learning, and conditional generation, can direct the generation process toward desired aims, regarding their biological activities, synthesis processes or chemical features. Given the recent emergence of these technologies and their relevance, this work presents a systematic and critical review on deep generative models and related optimization methods for targeted compound design, and their applications.

A language modeling-like approach to sketching

Sketching is a universal communication tool that, despite its simplicity, is able to efficiently express a large variety of concepts and, in some limited contexts, it can be even more immediate and effective than natural language. In this paper we explore the feasibility of using neural networks to approach sketching in the same way they are commonly used in Language Modeling. We propose a novel approach to what we refer to as "Sketch Modeling", in which a neural network is exploited to learn a probabilistic model that estimates the probability of sketches. We focus on simple sketches and, in particular, on the case in which sketches are represented as sequences of segments. Segments and sequences can be either given - when the sketches are originally drawn in this format - or automatically generated from the input drawing by means of a procedure that we designed to create short sequences, loosely inspired by the human behavior. A Recurrent Neural Network is used to learn the sketch model and, afterward, the network is seeded with an incomplete sketch that it is asked to complete, generating one segment at each time step. We propose a set of measures to evaluate the outcome of a Beam Search-based generation procedure, showing how they can be used to identify the most promising generations. Our experimental analysis assesses the feasibility of this way of modeling sketches, also in the case in which several different categories of sketches are considered.

DeepGRP: engineering a software tool for predicting genomic repetitive elements using Recurrent Neural Networks with attention

BACKGROUND

Repetitive elements contribute a large part of eukaryotic genomes. For example, about 40 to 50% of human, mouse and rat genomes are repetitive. So identifying and classifying repeats is an important step in genome annotation. This annotation step is traditionally performed using alignment based methods, either in a de novo approach or by aligning the genome sequence to a species specific set of repetitive sequences. Recently, Li (Bioinformatics 35:4408-4410, 2019) developed a novel software tool dna-brnn to annotate repetitive sequences using a recurrent neural network trained on sample annotations of repetitive elements.

RESULTS

We have developed the methods of dna-brnn further and engineered a new software tool DeepGRP. This combines the basic concepts of Li (Bioinformatics 35:4408-4410, 2019) with current techniques developed for neural machine translation, the attention mechanism, for the task of nucleotide-level annotation of repetitive elements. An evaluation on the human genome shows a 20% improvement of the Matthews correlation coefficient for the predictions delivered by DeepGRP, when compared to dna-brnn. DeepGRP predicts two additional classes of repeats (compared to dna-brnn) and is able to transfer repeat annotations, using RepeatMasker-based training data to a different species (mouse). Additionally, we could show that DeepGRP predicts repeats annotated in the Dfam database, but not annotated by RepeatMasker. DeepGRP is highly scalable due to its implementation in the TensorFlow framework. For example, the GPU-accelerated version of DeepGRP is approx. 1.8 times faster than dna-brnn, approx. 8.6 times faster than RepeatMasker and over 100 times faster than HMMER searching for models of the Dfam database.

CONCLUSIONS

By incorporating methods from neural machine translation, DeepGRP achieves a consistent improvement of the quality of the predictions compared to dna-brnn. Improved running times are obtained by employing TensorFlow as implementation framework and the use of GPUs. By incorporating two additional classes of repeats, DeepGRP provides more complete annotations, which were evaluated against three state-of-the-art tools for repeat annotation.

Deep recurrent model for individualized prediction of Alzheimer's disease progression

Alzheimer's disease (AD) is known as one of the major causes of dementia and is characterized by slow progression over several years, with no treatments or available medicines. In this regard, there have been efforts to identify the risk of developing AD in its earliest time. While many of the previous works considered cross-sectional analysis, more recent studies have focused on the diagnosis and prognosis of AD with longitudinal or time series data in a way of disease progression modeling. Under the same problem settings, in this work, we propose a novel computational framework that can predict the phenotypic measurements of MRI biomarkers and trajectories of clinical status along with cognitive scores at multiple future time points. However, in handling time series data, it generally faces many unexpected missing observations. In regard to such an unfavorable situation, we define a secondary problem of estimating those missing values and tackle it in a systematic way by taking account of temporal and multivariate relations inherent in time series data. Concretely, we propose a deep recurrent network that jointly tackles the four problems of (i) missing value imputation, (ii) phenotypic measurements forecasting, (iii) trajectory estimation of a cognitive score, and (iv) clinical status prediction of a subject based on his/her longitudinal imaging biomarkers. Notably, the learnable parameters of all the modules in our predictive models are trained in an end-to-end manner by taking the morphological features and cognitive scores as input, with our circumspectly defined loss function. In our experiments over The Alzheimers Disease Prediction Of Longitudinal Evolution (TADPOLE) challenge cohort, we measured performance for various metrics and compared our method to competing methods in the literature. Exhaustive analyses and ablation studies were also conducted to better confirm the effectiveness of our method.

An autonomous framework for interpretation of 3D objects geometric data using 2D images for application in additive manufacturing

Additive manufacturing, artificial intelligence and cloud manufacturing are three pillars of the emerging digitized industrial revolution, considered in industry 4.0. The literature shows that in industry 4.0, intelligent cloud based additive manufacturing plays a crucial role. Considering this, few studies have accomplished an integration of the intelligent additive manufacturing and the service oriented manufacturing paradigms. This is due to the lack of prerequisite frameworks to enable this integration. These frameworks should create an autonomous platform for cloud based service composition for additive manufacturing based on customer demands. One of the most important requirements of customer processing in autonomous manufacturing platforms is the interpretation of the product shape; as a result, accurate and automated shape interpretation plays an important role in this integration. Unfortunately despite this fact, accurate shape interpretation has not been a subject of research studies in the additive manufacturing, except limited studies aiming machine level production process. This paper has proposed a framework to interpret shapes, or their informative two dimensional pictures, automatically by decomposing them into simpler shapes which can be categorized easily based on provided training data. To do this, two algorithms which apply a Recurrent Neural Network and a two dimensional Convolutional Neural Network as decomposition and recognition tools respectively are proposed. These two algorithms are integrated and case studies are designed to demonstrate the capabilities of the proposed platform. The results suggest that considering the complex objects which can be decomposed with planes perpendicular to one axis of Cartesian coordination system and parallel withother two, the decomposition algorithm can even give results using an informative 2D image of the object.

Influence of medical domain knowledge on deep learning for Alzheimer's disease prediction

BACKGROUND AND OBJECTIVE

Alzheimer's disease (AD) is the most common type of dementia that can seriously affect a person's ability to perform daily activities. Estimates indicate that AD may rank third as a cause of death for older people, after heart disease and cancer. Identification of individuals at risk for developing AD is imperative for testing therapeutic interventions. The objective of the study was to determine could diagnostics of AD from EMR data alone (without relying on diagnostic imaging) be significantly improved by applying clinical domain knowledge in data preprocessing and positive dataset selection rather than setting naïve filters.

METHODS

Data were extracted from the repository of heterogeneous ambulatory EMR data, collected from primary care medical offices all over the U.S. Medical domain knowledge was applied to build a positive dataset from data relevant to AD. Selected Clinically Relevant Positive (SCRP) datasets were used as inputs to a Long-Short-Term Memory (LSTM) Recurrent Neural Network (RNN) deep learning model to predict will the patient develop AD.

RESULTS

Risk scores prediction of AD using the drugs domain information in an SCRP AD dataset of 2,324 patients achieved high out-of-sample score - 0.98-0.99 Area Under the Precision-Recall Curve (AUPRC) when using 90% of SCRP dataset for training. AUPRC dropped to 0.89 when training the model using less than 1,500 cases from the SCRP dataset. The model was still significantly better than when using naïve dataset selection.

CONCLUSION

The LSTM RNN method that used data relevant to AD performed significantly better when learning from the SCRP dataset than when datasets were selected naïvely. The integration of qualitative medical knowledge for dataset selection and deep learning technology provided a mechanism for significant improvement of AD prediction. Accurate and early prediction of AD is significant in the identification of patients for clinical trials, which can possibly result in the discovery of new drugs for treatments of AD. Also, the contribution of the proposed predictions of AD is a better selection of patients who need imaging diagnostics for differential diagnosis of AD from other degenerative brain disorders.

Online supervised attention-based recurrent depth estimation from monocular video

Autonomous driving highly depends on depth information for safe driving. Recently, major improvements have been taken towards improving both supervised and self-supervised methods for depth reconstruction. However, most of the current approaches focus on single frame depth estimation, where quality limit is hard to beat due to limitations of supervised learning of deep neural networks in general. One of the way to improve quality of existing methods is to utilize temporal information from frame sequences. In this paper, we study intelligent ways of integrating recurrent block in common supervised depth estimation pipeline. We propose a novel method, which takes advantage of the convolutional gated recurrent unit (convGRU) and convolutional long short-term memory (convLSTM). We compare use of convGRU and convLSTM blocks and determine the best model for real-time depth estimation task. We carefully study training strategy and provide new deep neural networks architectures for the task of depth estimation from monocular video using information from past frames based on attention mechanism. We demonstrate the efficiency of exploiting temporal information by comparing our best recurrent method with existing image-based and video-based solutions for monocular depth reconstruction.

ESTIMATING REPRODUCIBLE FUNCTIONAL NETWORKS ASSOCIATED WITH TASK DYNAMICS USING UNSUPERVISED LSTMS

We propose a method for estimating more reproducible functional networks that are more strongly associated with dynamic task activity by using recurrent neural networks with long short term memory (LSTMs). The LSTM model is trained in an unsupervised manner to learn to generate the functional magnetic resonance imaging (fMRI) time-series data in regions of interest. The learned functional networks can then be used for further analysis, e.g., correlation analysis to determine functional networks that are strongly associated with an fMRI task paradigm. We test our approach and compare to other methods for decomposing functional networks from fMRI activity on 2 related but separate datasets that employ a biological motion perception task. We demonstrate that the functional networks learned by the LSTM model are more strongly associated with the task activity and dynamics compared to other approaches. Furthermore, the patterns of network association are more closely replicated across subjects within the same dataset as well as across datasets. More reproducible functional networks are essential for better characterizing the neural correlates of a target task.

Randomized SMILES strings improve the quality of molecular generative models

Recurrent Neural Networks (RNNs) trained with a set of molecules represented as unique (canonical) SMILES strings, have shown the capacity to create large chemical spaces of valid and meaningful structures. Herein we perform an extensive benchmark on models trained with subsets of GDB-13 of different sizes (1 million, 10,000 and 1000), with different SMILES variants (canonical, randomized and DeepSMILES), with two different recurrent cell types (LSTM and GRU) and with different hyperparameter combinations. To guide the benchmarks new metrics were developed that define how well a model has generalized the training set. The generated chemical space is evaluated with respect to its uniformity, closedness and completeness. Results show that models that use LSTM cells trained with 1 million randomized SMILES, a non-unique molecular string representation, are able to generalize to larger chemical spaces than the other approaches and they represent more accurately the target chemical space. Specifically, a model was trained with randomized SMILES that was able to generate almost all molecules from GDB-13 with a quasi-uniform probability. Models trained with smaller samples show an even bigger improvement when trained with randomized SMILES models. Additionally, models were trained on molecules obtained from ChEMBL and illustrate again that training with randomized SMILES lead to models having a better representation of the drug-like chemical space. Namely, the model trained with randomized SMILES was able to generate at least double the amount of unique molecules with the same distribution of properties comparing to one trained with canonical SMILES.

Captioning Ultrasound Images Automatically

We describe an automatic natural language processing (NLP)-based image captioning method to describe fetal ultrasound video content by modelling the vocabulary commonly used by sonographers and sonologists. The generated captions are similar to the words spoken by a sonographer when describing the scan experience in terms of visual content and performed scanning actions. Using full-length second-trimester fetal ultrasound videos and text derived from accompanying expert voice-over audio recordings, we train deep learning models consisting of convolutional neural networks and recurrent neural networks in merged configurations to generate captions for ultrasound video frames. We evaluate different model architectures using established general metrics (BLEU, ROUGE-L) and application-specific metrics. Results show that the proposed models can learn joint representations of image and text to generate relevant and descriptive captions for anatomies, such as the spine, the abdomen, the heart, and the head, in clinical fetal ultrasound scans.

Magnetic Resonance Fingerprinting Reconstruction Using Recurrent Neural Networks

Magnetic Resonance Fingerprinting (MRF) is an imaging technique acquiring unique time signals for different tissues. Although the acquisition is highly accelerated, the reconstruction time remains a problem, as the state-of-the-art template matching compares every signal with a set of possible signals. To overcome this limitation, deep learning based approaches, e.g. Convolutional Neural Networks (CNNs) have been proposed. In this work, we investigate the applicability of Recurrent Neural Networks (RNNs) for this reconstruction problem, as the signals are correlated in time. Compared to previous methods based on CNNs, RNN models yield significantly improved results using in-vivo data.

A Model-Based Recurrent Neural Network With Randomness for Efficient Control With Applications

Recently, Recurrent Neural Network (RNN) control schemes for redundant manipulators have been extensively studied. These control schemes demonstrate superior computational efficiency, control precision, and control robustness. However, they lack planning completeness. This paper explains why RNN control schemes suffer from the problem. Based on the analysis, this work presents a new random RNN control scheme, which 1) introduces randomness into RNN to address the planning completeness problem, 2) improves control precision with a new optimization target, 3) improves planning efficiency through learning from exploration. Theoretical analyses are used to prove the global stability, the planning completeness, and the computational complexity of the proposed method. Software simulation is provided to demonstrate the improved robustness against noise, the planning completeness and the improved planning efficiency of the proposed method over benchmark RNN control schemes. Real-world experiments are presented to demonstrate the application of the proposed method.

A Novel Recurrent Neural Network for Improving Redundant Manipulator Motion Planning Completeness

Recurrent Neural Networks (RNNs) demonstrated advantages on control precision, system robustness and computational efficiency, and have been widely applied to redundant manipulator control optimization. Existing RNN control schemes locally optimize trajectories and are efficient and reliable on obstacle avoidance. However, for motion planning, they suffer from local minimum and do not have planning completeness. This work explained the cause of the planning incompleteness and addressed the problem with a novel RNN control scheme. The paper presented the proposed method in detail and analyzed the global stability and the planning completeness in theory. The proposed method was compared with other three control schemes on the precision, the robustness and the planning completeness in software simulation and the results shows the proposed method has improved precision and robustness, and planning completeness.