Multi-scale Pipeline Research Articles

Sampling globally and locally correct RNA 3D structures using Ernwin, SPQR and experimental SAXS data.

The determination of the three-dimensional structure of large RNA macromolecules in solution is a challenging task that often requires the use of several experimental and computational techniques. Small-angle X-ray scattering can provide insight into some geometrical properties of the probed molecule, but this data must be properly interpreted in order to generate a three-dimensional model. Here, we propose a multiscale pipeline which introduces SAXS data into modelling the global shape of RNA in solution, which can be hierarchically refined until reaching atomistic precision in explicit solvent. The low-resolution helix model (Ernwin) deals with the exploration of the huge conformational space making use of the SAXS data, while a nucleotide-level model (SPQR) removes clashes and disentangles the proposed structures, leading the structure to an all-atom representation in explicit water. We apply the procedure on four different known pdb structures up to 159 nucleotides with promising results. Additionally, we predict an all-atom structure for the Plasmodium falceparum signal recognition particle ALU RNA based on SAXS data deposited in the SASBDB, which has an alternate conformation and better fit to the SAXS data than the previously published structure based on the same data but other modelling methods.

Learning multiscale pipeline gated fusion for underwater image enhancement

Learning multiscale pipeline gated fusion for underwater image enhancement

Meta-analytic electric field modeling and developing multi-scale models

Abstract In this talk, I will present two recent developments of my lab using electric field modeling. First, I will talk about our efforts in combining electric field modeling with traditional meta-analysis on the effects of tDCS on working memory. Due to differences in electrode montages and stimulation intensities across different studies, results are difficult to aggregate for meta-analytic inferences. To overcome these limitations, we have developed a novel meta-analytic method relating behavioral effect sizes to electric field strength, to identify brain regions underlying the largest tDCS-induced WM improvement. Simulations on 69 studies targeting left prefrontal cortex showed that tDCS electric field strength in lower dorsolateral prefrontal cortex (Brodmann area 45/47) relates most strongly to improved WM performance. This brain region could be a target area for future tDCS studies. Our metanalytic framework can be applied to other stimulation modalities and behavioral measures. In the second part, I will talk about our efforts developing multi-scale models for TMS. While the induced electric field is the key factor to determine dosing and target engagement, it is only a first step to model the physiological response. To advance these efforts we have developed a new multi-scale pipeline (NeMo-TMS) for modeling TMS effects across spatial scales. On the macro-scale, we simulate TMS electric fields using SimNIBS. Afterwards, simulated electric fields from the previous step are coupled with morphologically realistic neuronal models in the NEURON environment. These neuron-scale simulations allow the investigation of membrane voltage (depolarization/hyperpolarization), action potential initiation and propagation, field intensity, and orientation necessary for modulating neuron response, etc. Finally, we incorporate the membrane voltage data to simulate the calcium concentration induced by voltage-gated calcium-channels at the subcellular scale by solving the calcium dynamics equations. This allows us to model effects of rTMS protocols on somatic calcium accumulation, important for neural plasticity.

Multiscale imaging of the rat brain using an integrated diceCT and histology workflow.

Advancements in tissue visualization techniques have spurred significant gains in the biomedical sciences by enabling researchers to integrate their datasets across anatomical scales. Of particular import are techniques that enable the interpolation of multiple hierarchical scales in samples taken from the same individuals. In this study, we demonstrate that two-dimensional histology techniques can be employed on neural tissues following three-dimensional diffusible iodine-based contrast-enhanced computed tomography (diceCT) without causing tissue degradation. This represents the first step toward a multiscale pipeline for brain visualization. We studied brains from adolescent male Sprague-Dawley rats, comparing experimental (diceCT-stained then de-stained) to control (without diceCT) brains to examine neural tissues for immunolabeling integrity, compare somata sizes, and distinguish neurons from glial cells within the telencephalon and diencephalon. We hypothesized that if experimental and control samples do not differ significantly in morphological cell analysis, then brain tissues are robust to the chemical, temperature, and radiation environments required for these multiple, successive imaging protocols. Visualizations for experimental brains were first captured via micro-computed tomography scanning of isolated, iodine-infused specimens. Samples were then cleared of iodine, serially sectioned, and prepared again using immunofluorescent, fluorescent, and cresyl violet labeling, followed by imaging with confocal and light microscopy, respectively. Our results show that many neural targets are resilient to diceCT imaging and compatible with downstream histological staining as part of a low-cost, multiscale brain imaging pipeline.

Two-stage multi-scale breast mass segmentation for full mammogram analysis without user intervention

Mammography is the primary imaging modality used for early detection and diagnosis of breast cancer. X-ray mammogram analysis mainly refers to the localization of suspicious regions of interest followed by segmentation, towards further lesion classification into benign versus malignant. Among diverse types of breast abnormalities, masses are the most important clinical findings of breast carcinomas. However, manually segmenting breast masses from native mammograms is time-consuming and error-prone. Therefore, an integrated computer-aided diagnosis system is required to assist clinicians for automatic and precise breast mass delineation. In this work, we present a two-stage multi-scale pipeline that provides accurate mass contours from high-resolution full mammograms. First, we propose an extended deep detector integrating a multi-scale fusion strategy for automated mass localization. Second, a convolutional encoder-decoder network using nested and dense skip connections is employed to fine-delineate candidate masses. Unlike most previous studies based on segmentation from regions, our framework handles mass segmentation from native full mammograms without any user intervention. Trained on INbreast and DDSM-CBIS public datasets, the pipeline achieves an overall average Dice of 80.44% on INbreast test images, outperforming state-of-the-art. Our system shows promising accuracy as an automatic full-image mass segmentation system. Extensive experiments reveals robustness against the diversity of size, shape and appearance of breast masses, towards better interaction-free computer-aided diagnosis.

A multi-scale pipeline linking drug transcriptomics with pharmacokinetics predicts in vivo interactions of tuberculosis drugs

Tuberculosis (TB) is the deadliest infectious disease worldwide. The design of new treatments for TB is hindered by the large number of candidate drugs, drug combinations, dosing choices, and complex pharmaco-kinetics/dynamics (PK/PD). Here we study the interplay of these factors in designing combination therapies by linking a machine-learning model, INDIGO-MTB, which predicts in vitro drug interactions using drug transcriptomics, with a multi-scale model of drug PK/PD and pathogen-immune interactions called GranSim. We calculate an in vivo drug interaction score (iDIS) from dynamics of drug diffusion, spatial distribution, and activity within lesions against various pathogen sub-populations. The iDIS of drug regimens evaluated against non-replicating bacteria significantly correlates with efficacy metrics from clinical trials. Our approach identifies mechanisms that can amplify synergistic or mitigate antagonistic drug interactions in vivo by modulating the relative distribution of drugs. Our mechanistic framework enables efficient evaluation of in vivo drug interactions and optimization of combination therapies.

High throughput automated detection of axial malformations in Medaka embryo

Fish embryo models are widely used as screening tools to assess the efficacy and/or toxicity of chemicals. This assessment involves the analysis of embryo morphological abnormalities. In this article, we propose a multi-scale pipeline to allow automated classification of fish embryos (Medaka: Oryzias latipes) based on the presence or absence of spine malformations. The proposed pipeline relies on the acquisition of fish embryo 2D images, on feature extraction based on mathematical morphology operators and on machine learning classification. After image acquisition, segmentation tools are used to detect the embryo before analysing several morphological features. An approach based on machine learning is then applied to these features to automatically classify embryos according to the presence of axial malformations. We built and validated our learning model on 1459 images with a 10-fold cross-validation by comparison with the gold standard of 3D observations performed under a microscope by a trained operator. Our pipeline results in correct classification in 85% of the cases included in the database. This percentage is similar to the percentage of success of a trained human operator working on 2D images. The key benefit of our approach is the low computational cost of our image analysis pipeline, which guarantees optimal throughput analysis.

Protein structural biology using cell-free platform from wheat germ

One of the biggest bottlenecks for structural analysis of proteins remains the creation of high-yield and high-purity samples of the target protein. Cell-free protein synthesis technologies are powerful and customizable platforms for obtaining functional proteins of interest in short timeframes, while avoiding potential toxicity issues and permitting high-throughput screening. These methods have benefited many areas of genomic and proteomics research, therapeutics, vaccine development and protein chip constructions. In this work, we demonstrate a versatile and multiscale eukaryotic wheat germ cell-free protein expression pipeline to generate functional proteins of different sizes from multiple host organism and DNA source origins. We also report on a robust purification procedure, which can produce highly pure (> 98%) proteins with no specialized equipment required and minimal time invested. This pipeline successfully produced and analyzed proteins in all three major geometry formats used for structural biology including single particle analysis with electron microscopy, and both two-dimensional and three-dimensional protein crystallography. The flexibility of the wheat germ system in combination with the multiscale pipeline described here provides a new workflow for rapid production and purification of samples that may not be amenable to other recombinant approaches for structural characterization.

A Multiscale pipeline for the search of string-induced CMB anisotropies

We propose a multi-scale edge-detection algorithm to search for the Gott-Kaiser-Stebbins imprints of a cosmic string (CS) network on the Cosmic Microwave Background (CMB) anisotropies. Curvelet decomposition and extended Canny algorithm are used to enhance the string detectability. Various statistical tools are then applied to quantify the deviation of CMB maps having a cosmic string contribution with respect to pure Gaussian anisotropies of inflationary origin. These statistical measures include the one-point probability density function, the weighted two-point correlation function (TPCF) of the anisotropies, the unweighted TPCF of the peaks and of the up-crossing map, as well as their cross-correlation. We use this algorithm on a hundred of simulated Nambu-Goto CMB flat sky maps, covering approximately $10\%$ of the sky, and for different string tensions $G\mu$. On noiseless sky maps with an angular resolution of $0.9'$, we show that our pipeline detects CSs with $G\mu$ as low as $G\mu\gtrsim 4.3\times 10^{-10}$. At the same resolution, but with a noise level typical to a CMB-S4 phase II experiment, the detection threshold would be to $G\mu\gtrsim 1.2 \times 10^{-7}$.

A multi-scale pipeline for reproducible analyses of tomato leaf expansion and its plasticity

Cell division and expansion both contribute to final leaf size in different plant species (Gonzalez et al., 2012). Correlations between final leaf size and cell number suggest that the cell cycle is a key process in leaf growth control. In some species, cells can also enter the specialized endoreduplication cycle. This variant of the cell cycle allows iterative DNA replication without cell division.

An automated assay for the assessment of cardiac arrest in fish embryo

Studies on fish embryo models are widely developed in research. They are used in several research fields including drug discovery or environmental toxicology. In this article, we propose an entirely automated assay to detect cardiac arrest in Medaka (Oryzias latipes) based on image analysis. We propose a multi-scale pipeline based on mathematical morphology. Starting from video sequences of entire wells in 24-well plates, we focus on the embryo, detect its heart, and ascertain whether or not the heart is beating based on intensity variation analysis. Our image analysis pipeline only uses commonly available operators. It has a low computational cost, allowing analysis at the same rate as acquisition. From an initial dataset of 3192 videos, 660 were discarded as unusable (20.7%), 655 of them correctly so (99.25%) and only 5 incorrectly so (0.75%). The 2532 remaining videos were used for our test. On these, 45 errors were made, leading to a success rate of 98.23%.

BiPPred: Combined sequence- and structure-based prediction of peptide binding to the Hsp70 chaperone BiP.

Substrate binding to Hsp70 chaperones is involved in many biological processes, and the identification of potential substrates is important for a comprehensive understanding of these events. We present a multi-scale pipeline for an accurate, yet efficient prediction of peptides binding to the Hsp70 chaperone BiP by combining sequence-based prediction with molecular docking and MMPBSA calculations. First, we measured the binding of 15mer peptides from known substrate proteins of BiP by peptide array (PA) experiments and performed an accuracy assessment of the PA data by fluorescence anisotropy studies. Several sequence-based prediction models were fitted using this and other peptide binding data. A structure-based position-specific scoring matrix (SB-PSSM) derived solely from structural modeling data forms the core of all models. The matrix elements are based on a combination of binding energy estimations, molecular dynamics simulations, and analysis of the BiP binding site, which led to new insights into the peptide binding specificities of the chaperone. Using this SB-PSSM, peptide binders could be predicted with high selectivity even without training of the model on experimental data. Additional training further increased the prediction accuracies. Subsequent molecular docking (DynaDock) and MMGBSA/MMPBSA-based binding affinity estimations for predicted binders allowed the identification of the correct binding mode of the peptides as well as the calculation of nearly quantitative binding affinities. The general concept behind the developed multi-scale pipeline can readily be applied to other protein-peptide complexes with linearly bound peptides, for which sufficient experimental binding data for the training of classical sequence-based prediction models is not available. Proteins 2016; 84:1390-1407. © 2016 Wiley Periodicals, Inc.

From Single Variants to Protein Cascades: MULTISCALE MODELING OF SINGLE NUCLEOTIDE VARIANT SETS IN GENETIC DISORDERS

Understanding the role of genetics in disease has become a central part of medical research. Non-synonymous single nucleotide variants (nsSNVs) in coding regions of human genes frequently lead to pathological phenotypes. Beyond single variations, the individual combination of nsSNVs may add to pathogenic processes. We developed a multiscale pipeline to systematically analyze the existence of quantitative effects of multiple nsSNVs and gene combinations in single individuals on pathogenicity. Based on this pipeline, we detected in a data set of 842 nsSNVs discovered in 76 genes related to cardiomyopathies, associated nsSNV combinations in seven genes present in at least 70% of all 639 patient samples, but not in a control cohort of healthy humans. Structural analyses of these revealed primarily an influence on the protein stability. For amino acid substitutions located at the protein surface, we generally observed a proximity to putative binding pockets. To computationally analyze cumulative effects and their impact, pathogenicity methods are currently being developed. Our approach supports this process, as shown on the example of a cardiac phenotype but can be likewise applied to other diseases such as cancer.

A medaka brain mutant, Ot (one-sided optic tectum), that affects symmetrical evagination of the developing optic tectum

Studies on fish embryo models are widely developed in research. They are used in several research fields including drug discovery or environmental toxicology. In this article, we propose an entirely automated assay to detect cardiac arrest in Medaka (Oryzias latipes) based on image analysis. We propose a multi-scale pipeline based on mathematical morphology. Starting from video sequences of entire wells in 24-well plates, we focus on the embryo, detect its heart, and ascertain whether or not the heart is beating based on intensity variation analysis. Our image analysis pipeline only uses commonly available operators. It has a low computational cost, allowing analysis at the same rate as acquisition. From an initial dataset of 3192 videos, 660 were discarded as unusable (20.7%), 655 of them correctly so (99.25%) and only 5 incorrectly so (0.75%). The 2532 remaining videos were used for our test. On these, 45 errors were made, leading to a success rate of 98.23%.