Recent Submissions

  • Stealthy Rootkit Attacks on Cyber-Physical Microgrids

    Rath, Suman; Zografopoulos, Ioannis; Konstantinou, Charalambos (2021-05-26) [Poster]
    Cyber-physical microgrids hold the key to a carbon-neutral power sector since they enable renewable and distributed energy resource integration, can alleviate overloaded distribution systems, and provide economic energy by generating and consuming power locally. The utilization of cyber-physical assets such as controllers, IoT sensors and actuators, and communication devices can enhance the stability and improve the control of microgrids. However, such assets, if maliciously operated, can become attack entry points and jeopardize the grid operation. Blind and uncoordinated cyber-attacks can be identified by existing security measures overcoming potential operational disruptions. However, rootkit attacks can stay hidden within cyber-physical systems and leverage system information to mask their presence. Rootkit detection is a strenuous process and requires advanced security methods due to their sophisticated operation. A careful analysis of possible rootkit target locations and their exploitation techniques is necessary to design effective threat detection and mitigation mechanisms. This paper discusses the cyber kill chain of a rootkit which can simultaneously deploy itself at multiple locations in a microgrid in a coordinated and stealthy way in order to maximize the impact on power system operations. The rootkit leverages system measurements to hide its presence and its attack impact from the detection mechanisms.
  • Improving Dielectric And Magnetic Properties Of (Cr, Fe, Ni)-Doped Sic Microwaves Absorbents: A Dft Study

    Merabet, Boualem (2021-05-24) [Poster]
    In wireless telecoms, EM waves absorbers become important if applied outside special fields like rooms, radar systems, and military application. Composite materials allow convenient use on surfaces, good control over mechanical properties, variation of EM properties with proper selection of matrix material and different inclusions. Wide frequency range, zero external magnetic field, thin absorption layer (required for absorbers) limit FM materials for microwave frequency range. In absorber composites, and FM inclusions reduce impedance mismatches at front interface of absorbers and increase absorption of EM waves (V. B. Bregar, IEEE Transactions on Magnetics (2004) 40, 3). Cr4+ transition metal ions provide a rich set of optically active defect spins in wide bandgap semiconductors, and produce in SiC a spin-1 ground state with a narrow, spectrally isolated, spin-selective, near-telecom optical interface (B. Diler, npj Quantum Inf (2020) 6, 11). Cr4+ are detected by placing the device into a photonic cavity to reduce the excited state lifetime by Purcell enhancement: large fraction of indistinguishable photons in near telecom ZPL would be further enhanced (A. M. Dibos et al., Phys. Rev. Lett. (2018) 120, 243601). A metallic character shown by (Ni, Cr)-codoped 4H–SiC, and a FM order mainly due to Cr impurities, originating from a strong FM coupling due to p-d hybridization interaction, allow us using our alloys in microwave circuits as absorbers. 4H-SiC (P63mc hexagonal structure of a =b =3.081 Å, c = 10.096 Å (B. Song, et al., J. Am. Chem. Soc. 131 (2009) 1376–1377)). To avoid EM pollution caused by electronic and telecom systems, Cr-doped 4H–SiC are used. Cr behaves as donor or acceptor and the dielectric properties of SiC can be changed through n- or p-type doping in the microwave range, where enhanced dielectric loss and improved EM matching are beneficial to get excellent microwave absorption performancee.(Justo JF, MachadoWVM, Assali LVC. Physica B 2006;378:376).
  • Effect Of Infill Density And Build Orientation On The Mechanical Properties Of Pla Cf & Petg Cf Composites In 3D Printing

    Said, Ahmed A.; Aljafari, Mutaz (2021-05-24) [Poster]
    Abstract An important aspect of additive manufacturing is reducing the weight while maintaining the mechanical properties of the material. While the properties of the additive manufactured specimen depend on the type of the materials used for the filament, the settings of printing also affect the properties of specimen. Consequently, this work is about the investigation of the effect of 3D printing settings on the weight reduction and the mechanical properties of the specimen between two different filament materials and to achieve an optimum compromise between the two. The infill density and the build orientation and pattren of the printed layers are the variables being considered while all other setting remain constant in all tests. Experiments are performed using two different filaments materials types; PLA Carbon fiber and PETG Carbon fiber composites. Ahmed Abdulrahman Said King Abdulaziz University Email: ahmad.a.alsayed@hotmail.com
  • Ultrasensitive Wireless Strain Sensor For Structural Health Monitoring

    Nesser, Hussein (2021-05-24) [Poster]
    Objective: Embedding a monitoring system including :sensor, cable, and interface circuit is a hindrance for some structures. Our project aims to develop new generations of strain sensors that are ultra-sensitive, have wireless communication of data and energy, low consumption of power, easy installation in-situ structures. Methods: The data and energy,  from and to LRC tag, will be transmitted remotely by an inductive coupling between the internal inductance of the sensor and an external readout coil. The external strain is detected by following the variation on the quality factor of the resonance frequency of the LRC circuit resulting from the variation of resistance of cracked electrodes. Results: -Microfabrication method, alternative to silicon technology, is used in the fabrication of our sensor. -Durable, flexible and thin materials like polyimide or PET, are used in the fabrication of our sensor which facilitates the process of integrating the sensor. -High crack density is created in a nonmetric metal electrodes (Cr/Au). -High-sensitivity (GF= 6657) and low strain detection (<0.1% strain). à Working towards following the resonance frequency by an external readout coil.
  • Strategy For Toughening Interfaces In Adhesively Bonded Composite Joints

    Tao, Ran (2021-05-24) [Poster]
    Objective: Surface patterning strategy was proposed to facilitate the formation of adhesive ligaments using pulsed CO2laser irradiation, in order to enhance fracture toughness and arrest the crack propagation within adhesively bonded composite joints. Methods: Arrest regions (highlighted in green in the video), with higher interfacial strength but lower fracture toughness than the uniformly laser ablated (LA) baseline surface, are alternatively placed on the top and bottom CFRP/adhesive interface. Mode I fracture toughness was assessed through the double cantilever beam (DCB) configuration. Results: The proposed patterning strategy is promising to trigger adhesive ligaments and hold the separating arms, promoting a R-curve-like response of the bonded joint, where the energy release rate keeps increasing as the crack propagates. The proposed patterning strategy shed light to the design of reliable and safety adhesively bonded CFRP joints.
  • Laser-Based Pre-Treatment Of Secondary Bonded Composite T-Joints For Improved Energy Dissipation

    Hashem, Mjed (2021-05-24) [Poster]
    Enhanced composite T-joint energy dissipation using laser patterning strategy M. Hashem, A. Wagih, G. Lubineau Introduction Internal aircraft structures such as ribs, spars and stringers are required to be connected and must be able to transfer loads to the skin. Metallic joints require bolts and rivets which cause weight penalties and stress concentration. Composite T-joints allows the connection of structural components while maintaining low weight and high toughness. Objectives â ¢Investigate the efficacy of a novel CO_2 laser pre-treatment on the toughness of CFRP T-joints. â ¢Understand the failure mechanisms in toughened CFRP T-joints considering different surface ply orientations.   Methods CFRP T-joints were manufactured by using unidirectional carbon fiber prepregs composed of toughened epoxy resin and carbon fibers. After a general peel-ply treatment, the adherents experienced CO_2 laser treatment. We applied laser treatment with two different energy, low and high, to create laser pattern of cleaning, LC, and ablation, LA, treatment as shown in Fig. 1.  Upon the laser treatments the stiffeners and skins were secondary bonded using an Araldite 420 A/B adhesive and mechanical tested using pull-off tests. The baseline joint, where peel ply treatment was applied is nominated as â PPâ . The laser patterned joint, where the alternative high and low laser power was applied, is nominated as â B5G5â . Results High surface roughness profile fluctuations were noticed at LC treated regimes, but low profiles were observed at LA regimes. SEM shows how the high fluence LA treatment exposed fully the fibers as compared to LC which minorly removed contaminants from the surface. PP baseline resulted in catastrophic failure at low extensions (4 mm). However, the laser treated T-joints showed progressive failure with improved toughness and extensions. The maximum load and extension for B5G5 joint reach 1712 N and 17.8 mm, respectively, compared to  805 N and 4 mm for PP joint. Discussion The enhancement observed for B5G5 T-joints was related with the creation of adhesive ligaments between the top and bottom adhesive layers (Fig. 5). The adhesive ligaments were generated at the transition between LA and LC treatment due to the difference in roughness between both treatments as shown in Fig. 2, which arrest the crack propagation at one interface allowing crack migration to the other interface. The effect of laser patterning was optimised when a 0° ply fiber direction was placed at the interface.   Owing to the crack migration, the ligament formation and breakage during testing, progressive failure occurred in B5G5 T-joints with larger improvements in the energy dissipation (toughness) reaching 12 times larger than the conventional PP treatment. Conclusions   â ¢Laser pre-treatment provided enhanced toughness and energy dissipation as compared to PP.   â ¢Laser patterning resulted in up to ~12x enhanced energy dissipation by the activation of non-local damage mechanisms which produced progressive failure indicating safter joints.
  • A Multiscale Computational Framework To Predict The Nonlinear Response Of Fibre-Reinforced Polymer Composites

    Ullah, Dr Zahur (2021-05-24) [Poster]
    Objective: Development of multi-scale computational framework for the prediction of nonlinear micro/meso response of the fibre-reinforced polymer (FRP) composites. Methods: The multi-scale computational framework provides the macroscopic constitutive behaviour of the structures based on its microscopically heterogeneous representative volume element (RVE). Two dominant damage mechanisms 1: Matrix plasticity (using pressure-dependent paraboloidal yield criterion). Fibre-matrix decohesion (using zero thickness cohesive interface elements). Yarns/fibres as transversely isotropic materials (calculation of fibre directions using potential flow analysis) 2. Generalised imposition of the RVE boundary conditions which allows convenient switching between displacement, traction and periodic boundary conditions 3. Adoption of hierarchic finite elements, which permits the use of arbitrary order of approximation 4. Implementation of the computational framework in an open-source finite element software MoFEM (Mesh Oriented Finite Element Method) and is designed to take advantage of the high-performance computing 5.
  • A Compact Thermofluidic Soft Actuator

    Chellattoan, Ragesh (2021-05-24) [Poster]
    A compact thermo-fluidicactuator Objective: Soft actuators producing large motion in a short time is mainly based on stretchable polymers actuated by pneumatic pressure. But such systems consist of bulky components like motor, pump/compressor, tubes and valves. Here, we developed a fast-responding large-amplitude soft actuator based on a liquid-gas phase transition, resulting in a compact system.    Methods: The required pressure is generated solely by the electrically induced phase transition of a fluid in a cavity. We pay special attention to design variables to improve the response time and propose a new design for the electrodes which are the most critical components.    Results: Our bending actuator produces large motion in less than 7 seconds using a low voltage source (less than 50V), which is much faster than previously reported soft actuator based on phase transition.
  • Smartphone-Based Single-Camera Stereo-Vision System

    Yu, Liping; Bekdullayev, Nurlat; Lubineau, Gilles (2021-05-24) [Poster]
    Stereo-digital image correlation technique using two synchronized industrial-grade cameras is extensively used for full-field 3D shape, displacement and deformation measurements. However, its use in resource-limited institutions and field-settings is inhibited by the need for relatively expensive, bulky and complicated experimental set-ups. To mitigate this problem, we established a cost-effective and ultra-portable smartphone-based stereo-digital image correlation system, which only uses a ubiquitous smartphone and an optical attachment. This optical attachment is composed of four planar mirrors and a 3D-printed mirror support, and can split the incoming scene into two sub-images, simulating a stereovision system using two virtual smartphones.
  • Design And Analysis Of Spoolable Reinforced Thermoplastic Pipes (Rtp) For On-Shore Oil And Gas Application

    A. Arafath, Abdul Rahim; Al-Ghamdi, Ali; Khandelwal, Ratnesh (2021-05-24) [Poster]
    WHAT IS RTP? The term refers to a multilayer pipe construction where at least one layer is acting as reinforcement. APPLICATION OVERVIEW & REQUIREMENTS ONSHORE TESTING & VALIDATION –HOW CAN WE SHORTEN TIME TO MARKET? COMPOSITE LAY-UP OPTIMIZATION COMPOSITE FAILURE CRITERIA Tsai-Hill Failure Criterion (TH) Hashin’sfailure criteria (HF) Fiber failure criteria (FF) ANALYSIS OF BURST PRESSURE ANALYSIS OF MINIMUM BENDING RADIUS ANALYSIS OF LONG-TERM DESIGN LIFE RESULTS
  • The Future Of Nonmetallic Composite Materials In Upstream Applications

    Badeghaish, Wael; Noui-Mehidi, Mohamed; Salazar, Oscar (2021-05-24) [Poster]
  • Planning and development of a long-term digital self-management tool for osteoarthritis: the Intelligent Knee Osteoarthritis Lifestyle App

    Chowdhury, Enhad A; Ceballos Inza, Victor; Western, Max J; Walsh, Nicola E; Bilzon, James L J; Jones, Simon L (Rheumatology, Oxford University Press (OUP), 2021-04-26) [Poster]
    There is strong evidence for beneficial effects of physical activity for people with knee osteoarthritis (KOA). While supervised exercise programmes are effective, they are resource intensive, typically of limited duration and hard to implement at scale. More accessible options should be developed to enable individuals to adopt and maintain appropriate physically active lifestyles. Smartphone apps can monitor activity and symptoms, providing feedback to support self-management. We aimed to co-design a KOA self-management app, with physiotherapists and people experiencing KOA, to support long-term physical activity.
  • Abstraction Layer For Standardizing APIs of Task-Based Engines

    Alomairy, Rabab; Ltaief, Hatem; Abduljabbar, Mustafa; Keyes, David E. (2020-06-23) [Poster]
    AL4SAN is a lightweight library for abstracting the APIs of task-based runtime engines. AL4SAN unifies the expression of tasks and their data dependencies. It supports various dynamic runtime systems relying on compiler technology and user-defined APIs. It enables a single application to employ different runtimes and their respective scheduling components, while providing user-obliviousness to the underlying hardware configurations. AL4SAN exposes common front-end APIs and connects to different backend runtimes. AL4SAN enables runtime interoperability by switching runtimes at runtime. Blending runtime systems permits to achieve a twofold speedup on a task-based generalized symmetric eigenvalue solver, relative to state-ofthe- art implementations. The ultimate goal of AL4SAN is not to create a new runtime, but to strengthen co-design of existing runtimes/applications, while facilitating user productivity and code portability.
  • Lateral migration patterns toward or away from injection wells for earthquake clusters in Oklahoma

    Lopez Comino, Jose Angel; Galis, Martin; Mai, Paul Martin; Chen, Xiaowei; Stich, Daniel (Copernicus GmbH, 2020-03-09) [Poster]
    Exploring the connections between injection wells and seismic migration patterns is key to understanding processes controlling growth of fluid-injection induced seismicity. Numerous seismic clusters in Oklahoma have been associated with wastewater disposal operations, providing a unique opportunity to investigate migration directions of each cluster with respect to the injection-well locations. We introduce new directivity migration parameters to identify and quantify lateral migration toward or away from the injection wells. We take into account cumulative volume and injection rate from multiple injection wells. Our results suggest a weak relationship between migration direction and the cluster-well distances. Migration away from injection wells is found for distances shorter than 5-13 km, while an opposite migration towards the wells is observed for larger distances, suggesting an increasing influence of poroelastic stress changes. This finding is more stable when considering cumulative injected volume instead of injection rate. We do not observe any relationship between migration direction and injected volume or equivalent magnitudes.
  • A novel iPSC model of isogenic knockout of entire WAS gene can recapitulate WAS phenotypes in iPSC derived macrophages

    Yuan, Baolei; Zhou, Xuan; Ramos-Mandujano, Gerardo; V. Cortes-Medina, Lorena; Bi, Chongwei; Li, Mo (2020-1-20) [Poster]
    A novel iPSC model of isogenic knockout of entire WAS gene can recapitulate WAS phenotypes in iPSC derived macrophages Baolei Yuan1, Xuan Zhou1, Gerardo Ramos-Mandujano1, Lorena V. Cortes-Medina2, Chongwei Bi1, Mo Li1 Wiskott-Aldrich syndrome (WAS) is an X-linked recessive disease caused by mutations in the WAS protein (WASP). WAS is associated with devastating symptoms including microthrombocytopenia, eczema, autoimmunity and cancer. The molecular mechanism underlying WAS remains elusive thus far. The genotype-phenotype relationship in WAS is complex. There are over 200 mutations that lead to hypomorphic levels or complete loss of WASP, while it is impossible to predict clinical severity based on the mutation alone. To help evaluate phenotype variability due to mutational background of different patients, we developed an isogenic WASP-knockout (WASP-KO) induced pluripotent stem cell (iPSC) model using the CRISPR/Cas9 technique that completely removed the WAS gene. The isogenic iPSC model was differentiated into macrophages, which are reported to be affected by WASP mutations. This model can be used for studying the WASP functions, WAS disease mechanism and drug screening.
  • Deetal-Perio: DEEp denTAL Advisor for Periodontitis Diagnosis based on Two-step Segmentation of Teeth and Gingiva with Lower-dimensional Features

    Zhou, Juexiao; Li, Haoyang; Gao, Xin (2020-1-20) [Poster]
    Deetal-Perio: DEEp denTAL Advisor for Periodontitis Diagnosis based on Two-step Segmentation of Teeth and Gingiva with Lower-dimensional Features Haoyang Li1,2, Juexiao Zhou1,3 , Xin Gao1,* 1 Computational Bioscience Research Center (CBRC), Computer, Electrical, and Mathematical Sciences and Engineering (CEMSE) Division, King Abdullah University of Science and Technology (KAUST), Thuwal 23955, Saudi Arabia 2 MOE Key Laboratory of Symbolic Computation and Knowledge Engineering, College of Computer Science and Technology, Jilin University, Changchun 130012, China 3 Department of Biology, Southern University of Science and Technology, Shenzhen 518055, China   Background Periodontitis is often known as Gum Disease and is a very common condition in which the gums and deeper periodontal structures become inflamed. This inflammation is the result of response to the invasion bacteria influenced by genetic and lifestyle-associated factors1. Periodontitis usually takes the form of redness, swelling and a tendency to bleed during tooth brushing and the severe periodontitis ranks sixth in the Global Burden of Disease study that affects 11% of the world population2. Also, periodontitis may be a risk factor for cardiovascular disease3 and has an additive effect on development of diabetic complications4. X-ray is a widely used, economy and convenient method to scan the teeth and study the periodontal diseases. Therefore, the prediction of periodontitis based on X-ray image has high practical application value.   Highlights Lower-dimensional and interpretable features. Outperforms other state-of-the-art methods. Reveals the significance of crown-root ratio(CR) as the key feature for periodontitis prediction   Introduction The majority of the previous works on the prediction of periodontitis focus on mainly two categories of methods, traditional machine learning methods and CNN based methods, while the general form of input data are the raw image or multi-modal data of patients. Methods In this project, we predict the class of periodontitis based on X-ray images of patients following two-step segmentation of tooth and gingiva. • DatasetX-ray images of 300 patients are from dental clinics in China. The contour of teeth, gingiva and the level of periodontitis are annotated by professional dentists. • Segmentation of Teeth and GingivaThe segmentation of teeth and gingiva is based on our well-trained Mask-RCNN model. • Prediction and Calibration of Tooth Numbering The teeth numbering is predicted by both the multi-class Mask-RCNN (exact teeth numbering in the FDI numbering system) and binary Mask-RCNN (is a tooth or not). Then our calibration method will output the final teeth numbering results by integrating the results of both types of Mask-RCNN. • Calculation of ABL (Feature of Periodontitis) After the segmentation of teeth and gingiva, for each tooth, the loss of alveolar bone (ABL) is calculated with the largest perpendicular distance of both teeth crown and teeth root to the intersected gingiva. The 32 teeth of each sample will be reorganized into a 1x32 vector for the prediction of periodontitis. • Prediction of PeriodontitisThe 1x32 vector of teeth ratio is post-processed with interpolation, then the Synthetic Minority Oversampling (SMOTE) is adopted to solve the class-imbalance issue. Next, the XGboost is applied to do the classification of periodontitis. • Evaluation of MethodsMean average precision (mAP), Dice coefficient, Accuracy and F1-Score are used to evaluate our results. Results• Our method is powerful for teeth segmentation and numbering • Our method can handle both 3-Classes and 4-Classes classification and outperforms other compare methods • Our method is robust with respect to the class size References 1. Page RC, Kornman KS. The pathogenesis of human periodontitis: An introduction. Periodontol 2000 1997; 14: 9–11.2. Marcenes W, Kassebaum NJ, Bernabé E, et al. Global burden of oral conditions in 1990-2010: A systematic analysis. J Dent Res2013; 92: 592–597.3. Tonetti MS, Van Dyke TE; Working Group 1 of the Joint EFP/AAP Workshop. Periodontitis and atherosclerotic cardiovascular disease: Consensus report of the Joint European Federation of Periodontology and the American Academy of Periodontology Workshop on periodontitis and systemic diseases. J Clin Periodontol 2013; 40(Suppl. 14): S24–S29. 4. Lalla E, Papapanou PN. Diabetes mellitus and periodontitis: A tale of two common interrelated diseases. Nat Rev Endocrinol 2011; 7: 738–748.
  • Novel Feature Generation for Multiple Hand Gestures Classification

    Chahid, Abderrazak; Khushaba, Rami; Al-Jumaily, Adel; Laleg-Kirati, Taous-Meriem (2020-1-20) [Poster]
    Novel Feature Generation for Multiple Hand Gestures Classification Abderrazak Chahid 1, Rami Khushaba 2, Adel Al-Jumaily 2 and Taous-Meriem Laleg-Kirati 1 1 King Abdullah University of Science and Technology (KAUST).  2 University of Technology, Sydney (UTS), Australia   Abstract Surface electromyography (sEMG) signals represent an opportunity to control a multifunctional prosthetic hand in a non-invasive way. In this work, we investigate a novel feature extraction method that improves the interpretation of sEMG signal of multiple hand gestures.  So, missing body parts could be perfectly restored!!   Introduction   Since prosthesis invention, several prostheses were proposed to replace a missing body part, which may be lost through trauma, diseases, etc. Some of these solutions use the non-invasive sEMG signals to control this device 1,2.   Objective: - Build a smart prosthetic hand using artificial intelligence (AI) techniques. - Develop a generalizable and robust  AI model for multiple hand gesture’ predictions using a novel feature extraction method.   Challenges: - Some hand gesture have similar sEMG signals, - Prosthesis response in Real-Time and low cost.     Framework The proposed framework is described as follows: - Quantization:  sEMG signals are converted into sequences using a uniform Quantizer - QuPWM features: different features are extracted based on the Position Weight Matrix (PWM) method using multiple patterns (k-mers) 4.  - Classification: the extracted features are fed to standard classifiers for hand gestures classification.   Conclusion   - We developed a new feature extraction method using Quantization-based PWM (QuPWM) method. - The obtained results are very encouraging and with high accuracy for different subjects. - We believe that signal processing is a key to extract the inherent features from biomedical signals such as sEMG,…etc. - The proposed features will enhance human–computer interaction (HCI).   Future work  - Extensive validation using more dataset, - Combine these features with deep learning classifier to deal with big data, - Integrate the QuPWM in clinical practice: prosthesis.   References 1 Ciancio AL, Cordella F, Hoffmann KP, Schneider A, Guglielmelli E, Zollo L. Current achievements and future directions of hand prostheses controlled via peripheral nervous system. InThe Hand 2017 (pp. 75-95). Springer, Cham. 2 Ahsan MR, Ibrahimy MI, Khalifa OO. Electromygraphy (EMG) signal based hand gesture recognition using artificial neural network (ANN). In2011 4th International Conference on Mechatronics (ICOM) 2011 May 17 (pp. 1-6). IEEE. 3 Du Y, Wenguang J, Wentao W, Geng W. CapgMyo: a high density surface electromyography database for gesture recognition. 4 Chahid A, Albalawi F, Alotaiby TN, Al-Hameed MH, Alshebeili S, Laleg-Kirati TM. QuPWM: Feature Extraction Method for MEG Epileptic Spike Detection. Under revision in IEEE Journal of Biomedical and Health Informatics, arXiv preprint arXiv:1907.02596. 2019 Jul 3.    
  • NNfold: RNA secondary structure prediction by deep learning

    Umarov, Ramzan; Li, Yu; Van Neste, Christophe (2020-1-20) [Poster]
    NNfold: RNA secondary structure prediction by deep learning RNA molecules have a plethora of functions within the cell. These functions can be divided into information-carrier, catalytic, or structural (scaffolding of other molecules), or a combination. For the catalytic or regulation functionality the structure that the RNA molecule has is pivotal and predicting to which structure it is most likely to fold is therefore essential to fully understand its biological role. In general, RNA affects extensively protein regulation, through its control of gene expression, post-transcriptional modifications, or translational regulation. RNA secondary structure can be obtained by techniques such as X-ray diffraction and NMR. However, biological experimental methods are still inefficient and expensive. Thus, computational prediction algorithms are still widely used for predicting RNA secondary structures. Taking the raw sequence represented in a string, we first use a one-hot encoding. The encoded matrix has a dimension of L by 4. Then, the encoding will go through two models,  the local model and the global model, to extract local contact information and global contact information, respectively. Regrading the local model, the input for the model are two chunks of the raw encoding, whose dimensions are 20 by 20. Then we concatenate those 20 by 20 chunk matrices into the L by L local contact information matrix. We used six 1D convolutional layers and one fully-connected layer to model the local information.  In terms of the global model, we use three 1D convolutional layers to predict whether a base can pair with any other base or not, whose output is a vector of length L.  In the vector, 1 means the corresponding base may pair with the other base and 0 means that the corresponding base does not pair with this base.  To combine the local information and the global information, we convert the global vector into a symmetric matrix of L by L and perform a pairwise multiplication between the global information and the local information,  enforcing the global constraint into the preliminary contact map. After combining the global information and the local information, the obtained global contact map may still violate the two constraints mentioned above.  We used the following greedy sorting algorithm to resolve the conflict. We introduce NNfold, a sequence based deep learning method to predict RNA secondary structure. The predictions are made in two steps: first we construct a matrix with likelihood of each nucleotide pairing by predicting all potential interactions using convolutional deep learning model. Next, we modify the base pairs list obtained from the matrix using second model whose output is used to ensure validity of the final secondary structure. NNfold performed much better than thermodynamics-based methods on the diverse set of RNA sequences, improving average F1 score by 0.20. It is also capable of predicting pseudoknots which is a challenging task for other approaches.

View more