High-confidence 3D template matching for cryo-electron tomography.

Visual proteomics attempts to build atlases of the molecular content of cells but the automated annotation of cryo electron tomograms remains challenging. Template matching (TM) and methods based on machine learning detect structural signatures of macromolecules. However, their applicability remains limited in terms of both the abundance and size of the molecular targets. Here we show that the performance of TM is greatly improved by using template-specific search parameter optimization and by including higher-resolution information. We establish a TM pipeline with systematically tuned parameters for the automated, objective and comprehensive identification of structures with confidence 10 to 100-fold above the noise level. We demonstrate high-fidelity and high-confidence localizations of nuclear pore complexes, vaults, ribosomes, proteasomes, fatty acid synthases, lipid membranes and microtubules, and individual subunits inside crowded eukaryotic cells. We provide software tools for the generic implementation of our method that is broadly applicable towards realizing visual proteomics.

Transfer Learning Allows Accurate RBP Target Site Prediction with Limited Sample Sizes.

RNA-binding proteins are vital regulators in numerous biological processes. Their disfunction can result in diverse diseases, such as cancer or neurodegenerative disorders, making the prediction of their binding sites of high importance. Deep learning (DL) has brought about a revolution in various biological domains, including the field of protein-RNA interactions. Nonetheless, several challenges persist, such as the limited availability of experimentally validated binding sites to train well-performing DL models for the majority of proteins. Here, we present a novel training approach based on transfer learning (TL) to address the issue of limited data. Employing a sophisticated and interpretable architecture, we compare the performance of our method trained using two distinct approaches: training from scratch (SCR) and utilizing TL. Additionally, we benchmark our results against the current state-of-the-art methods. Furthermore, we tackle the challenges associated with selecting appropriate input features and determining optimal interval sizes. Our results show that TL enhances model performance, particularly in datasets with minimal training data, where satisfactory results can be achieved with just a few hundred RNA binding sites. Moreover, we demonstrate that integrating both sequence and evolutionary conservation information leads to superior performance. Additionally, we showcase how incorporating an attention layer into the model facilitates the interpretation of predictions within a biologically relevant context.

ENNGene: an Easy Neural Network model building tool for Genomics.

BACKGROUND: The recent big data revolution in Genomics, coupled with the emergence of Deep Learning as a set of powerful machine learning methods, has shifted the standard practices of machine learning for Genomics. Even though Deep Learning methods such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are becoming widespread in Genomics, developing and training such models is outside the ability of most researchers in the field. RESULTS: Here we present ENNGene-Easy Neural Network model building tool for Genomics. This tool simplifies training of custom CNN or hybrid CNN-RNN models on genomic data via an easy-to-use Graphical User Interface. ENNGene allows multiple input branches, including sequence, evolutionary conservation, and secondary structure, and performs all the necessary preprocessing steps, allowing simple input such as genomic coordinates. The network architecture is selected and fully customized by the user, from the number and types of the layers to each layer's precise set-up. ENNGene then deals with all steps of training and evaluation of the model, exporting valuable metrics such as multi-class ROC and precision-recall curve plots or TensorBoard log files. To facilitate interpretation of the predicted results, we deploy Integrated Gradients, providing the user with a graphical representation of an attribution level of each input position. To showcase the usage of ENNGene, we train multiple models on the RBP24 dataset, quickly reaching the state of the art while improving the performance on more than half of the proteins by including the evolutionary conservation score and tuning the network per protein. CONCLUSIONS: As the role of DL in big data analysis in the near future is indisputable, it is important to make it available for a broader range of researchers. We believe that an easy-to-use tool such as ENNGene can allow Genomics researchers without a background in Computational Sciences to harness the power of DL to gain better insights into and extract important information from the large amounts of data available in the field.

A deep learning framework for autonomous detection and classification of Crohn's disease lesions in the small bowel and colon with capsule endoscopy.

Background and study aims Small bowel ulcerations are efficiently detected with deep learning techniques, whereas the ability to diagnose Crohn's disease (CD) in the colon with it is unknown. This study examined the ability of a deep learning framework to detect CD lesions with pan-enteric capsule endoscopy (CE) and classify lesions of different severity. Patients and methods CEs from patients with suspected or known CD were included in the analysis. Two experienced gastroenterologists classified anonymized images into normal mucosa, non-ulcerated inflammation, aphthous ulceration, ulcer, or fissure/extensive ulceration. An automated framework incorporating multiple ResNet-50 architectures was trained. To improve its robustness and ability to characterize lesions, image processing methods focused on texture enhancement were employed. Results A total of 7744 images from 38 patients with CD were collected (small bowel 4972, colon 2772) of which 2748 contained at least one ulceration (small bowel 1857, colon 891). With a patient-dependent split of images for training, validation, and testing, ulcerations were diagnosed with a sensitivity, specificity, and diagnostic accuracy of 95.7 % (CI 93.4-97.4), 99.8 % (CI 99.2-100), and 98.4 % (CI 97.6-99.0), respectively. The diagnostic accuracy was 98.5 % (CI 97.5-99.2) for the small bowel and 98.1 % (CI 96.3-99.2) for the colon. Ulcerations of different severities were classified with substantial agreement (κ = 0.72). Conclusions Our proposed framework is in excellent agreement with the clinical standard, and diagnostic accuracies are equally high for the small bowel and colon. Deep learning approaches have a great potential to help clinicians detect, localize, and determine the severity of CD with pan-enteric CE.

Facial erythema detects diabetic neuropathy using the fusion of machine learning, random matrix theory and self organized criticality.

Rubeosis faciei diabeticorum, caused by microangiopathy and characterized by a chronic facial erythema, is associated with diabetic neuropathy. In clinical practice, facial erythema of patients with diabetes is evaluated based on subjective observations of visible redness, which often goes unnoticed leading to microangiopathic complications. To address this major shortcoming, we designed a contactless, non-invasive diagnostic point-of-care-device (POCD) consisting of a digital camera and a screen. Our solution relies on (1) recording videos of subject's face (2) applying Eulerian video magnification to videos to reveal important subtle color changes in subject's skin that fall outside human visual limits (3) obtaining spatio-temporal tensor expression profile of these variations (4) studying empirical spectral density (ESD) function of the largest eigenvalues of the tensors using random matrix theory (5) quantifying ESD functions by modeling the tails and decay rates using power law in systems exhibiting self-organized-criticality and (6) designing an optimal ensemble of learners to classify subjects into those with diabetic neuropathy and those of a control group. By analyzing a short video, we obtained a sensitivity of 100% in detecting subjects diagnosed with diabetic neuropathy. Our POCD paves the way towards the development of an inexpensive home-based solution for early detection of diabetic neuropathy and its associated complications.

Non-invasive detection of diabetic complications via pattern analysis of temporal facial colour variations.

BACKGROUND AND OBJECTIVE: Diabetes mellitus is a common disorder amounting to 400 million patients worldwide. It is often accompanied by a number of complications, including neuropathy, nephropathy, and cardiovascular diseases. For example, peripheral neuropathy is present among 20-30% of diabetics before the diagnosis is substantiated. For this reason, a reliable detection method for diabetic complications is crucial and attracts a lot of research attention. METHODS: In this paper, we introduce a non-invasive detection framework for patients with diabetic complications that only requires short video recordings of faces from a standard commercial camera. We employed multiple image processing and pattern recognition techniques to process video frames, extract relevant information, and predict the health status. To evaluate our framework, we collected a dataset of 114 video files from diabetic patients, who were diagnosed with diabetes for years and 60 video files from the control group. Extracted features from videos were tested using two conceptually different classifiers. RESULTS: We found that our proposed framework correctly identifies patients with diabetic complications with 92.86% accuracy, 100% sensitivity, and 80% specificity. CONCLUSIONS: Our study brings a novel perspective on diagnosis procedures in this field. We used multiple techniques from image processing, pattern recognition, and machine learning to robustly process video frames and predict the health status of our subjects with high efficiency.

Invisible Color Variations of Facial Erythema: A Novel Early Marker for Diabetic Complications?

AIM: (1) To quantify the invisible variations of facial erythema that occur as the blood flows in and out of the face of diabetic patients, during the blood pulse wave using an innovative image processing method, on videos recorded with a conventional digital camera and (2) to determine whether this "unveiled" facial red coloration and its periodic variations present specific characteristics in diabetic patients different from those in control subjects. METHODS: We video recorded the faces of 20 diabetic patients with peripheral neuropathy, retinopathy, and/or nephropathy and 10 nondiabetic control subjects, using a Canon EOS camera, for 240 s. Only one participant presented visible facial erythema. We applied novel image processing methods to make the facial redness and its variations visible and automatically detected and extracted the redness intensity of eight facial patches, from each frame. We compared average and standard deviations of redness in the two groups using t-tests. RESULTS: Facial redness varies, imperceptibly and periodically, between redder and paler, following the heart pulsation. This variation is consistently and significantly larger in diabetic patients compared to controls (p value < 0.001). CONCLUSIONS: Our study and its results (i.e., larger variations of facial redness with the heartbeats in diabetic patients) are unprecedented. One limitation is the sample size. Confirmation in a larger study would ground the development of a noninvasive cost-effective automatic tool for early detection of diabetic complications, based on measuring invisible redness variations, by image processing of facial videos captured at home with the patient's smartphone.

Validation of electron microscopy initial models via small angle X-ray scattering curves.

MOTIVATION: Cryo electron microscopy (EM) is currently one of the main tools to reveal the structural information of biological macromolecules. The re-construction of three-dimensional (3D) maps is typically carried out following an iterative process that requires an initial estimation of the 3D map to be refined in subsequent steps. Therefore, its determination is key in the quality of the final results, and there are cases in which it is still an open issue in single particle analysis (SPA). Small angle X-ray scattering (SAXS) is a well-known technique applied to structural biology. It is useful from small nanostructures up to macromolecular ensembles for its ability to obtain low resolution information of the biological sample measuring its X-ray scattering curve. These curves, together with further analysis, are able to yield information on the sizes, shapes and structures of the analyzed particles. RESULTS: In this paper, we show how the low resolution structural information revealed by SAXS is very useful for the validation of EM initial 3D models in SPA, helping the following refinement process to obtain more accurate 3D structures. For this purpose, we approximate the initial map by pseudo-atoms and predict the SAXS curve expected for this pseudo-atomic structure. The match between the predicted and experimental SAXS curves is considered as a good sign of the correctness of the EM initial map. AVAILABILITY AND IMPLEMENTATION: The algorithm is freely available as part of the Scipion 1.2 software at http://scipion.i2pc.es/.