Automated subcellular location determination and high-throughput microscopy

Bioimage Classification with Handcrafted and Learned Features

Bioimage classification is increasingly becoming more important in many biological studies including those that require accurate cell phenotype recognition, subcellular localization, and histopathological classification. In this paper, we present a new General Purpose (GenP) bioimage classification method that can be applied to a large range of classification problems. The GenP system we propose is an ensemble that combines multiple texture features (both handcrafted and learned descriptors) for superior and generalizable discriminative power. Our ensemble obtains a boosting of performance by combining local features, dense sampling features, and deep learning features. Each descriptor is used to train a different Support Vector Machine that is then combined by sum rule. We evaluate our method on a diverse set of bioimage classification tasks each represented by a benchmark database, including some of those available in the IICBU 2008 database. Each bioimage classification task represents a typical subcellular, cellular, and tissue level classification problem. Our evaluation on these datasets demonstrates that the proposed GenP bioimage ensemble obtains state-of-the-art performance without any ad-hoc dataset tuning of the parameters (thereby avoiding any risk of overfitting/overtraining). To reproduce the experiments reported in this paper, the MATLAB code of all the descriptors is available at https://github.com/LorisNanni and https://www.dropbox.com/s/bguw035yrqz0pwp/ElencoCode.docx?dl=0.

Rice phytoglobins regulate responses under low mineral nutrients and abiotic stresses in Arabidopsis thaliana

Just like animals, plants also contain haemoglobins (known as phytoglobins in plants). Plant phytoglobins (Pgbs) have been categorized into 6 different classes, namely, Phytogb0 (Pgb0), Phytogb1 (Pgb1), Phytogb2 (Pgb2), SymPhytogb (sPgb), Leghaemoglobin (Lb), and Phytogb3 (Pgb3). Among the 6 Phytogbs, sPgb and Lb have been functionally characterized, whereas understanding of the roles of other Pgbs is still evolving. In our present study, we have explored the function of 2 rice Pgbs (OsPgb1.1 and OsPgb1.2). OsPgb1.1, OsPgb1.2, OsPgb1.3, and OsPgb1.4 displayed increased level of transcript upon salt, drought, cold, and ABA treatment. The overexpression (OX) lines of OsPgb1.2 in Arabidopsis showed a tolerant phenotype in terms of better root growth in low potassium (K+ ) conditions. The expression of the known K+ gene markers such as LOX2, HAK5, and CAX3 was much higher in the OsPgb1.2 OX as compared to wild type. Furthermore, the OsPgb1.2 OX lines showed a decrease in reactive oxygen species (ROS) production and conversely an increase in the K+ content, both in root and shoot, as compared to wild type in K+ limiting condition. Our results indicated the potential involvement of OsPgb1.2 in signalling networks triggered by the nutrient deficiency stresses.

HPSLPred: An Ensemble Multi-Label Classifier for Human Protein Subcellular Location Prediction with Imbalanced Source

Predicting the subcellular localization of proteins is an important and challenging problem. Traditional experimental approaches are often expensive and time-consuming. Consequently, a growing number of research efforts employ a series of machine learning approaches to predict the subcellular location of proteins. There are two main challenges among the state-of-the-art prediction methods. First, most of the existing techniques are designed to deal with multi-class rather than multi-label classification, which ignores connections between multiple labels. In reality, multiple locations of particular proteins imply that there are vital and unique biological significances that deserve special focus and cannot be ignored. Second, techniques for handling imbalanced data in multi-label classification problems are necessary, but never employed. For solving these two issues, we have developed an ensemble multi-label classifier called HPSLPred, which can be applied for multi-label classification with an imbalanced protein source. For convenience, a user-friendly webserver has been established at http://server.malab.cn/HPSLPred.

pLoc-mGneg: Predict subcellular localization of Gram-negative bacterial proteins by deep gene ontology learning via general PseAAC

Information of the proteins' subcellular localization is crucially important for revealing their biological functions in a cell, the basic unit of life. With the avalanche of protein sequences generated in the postgenomic age, it is highly desired to develop computational tools for timely identifying their subcellular locations based on the sequence information alone. The current study is focused on the Gram-negative bacterial proteins. Although considerable efforts have been made in protein subcellular prediction, the problem is far from being solved yet. This is because mounting evidences have indicated that many Gram-negative bacterial proteins exist in two or more location sites. Unfortunately, most existing methods can be used to deal with single-location proteins only. Actually, proteins with multi-locations may have some special biological functions important for both basic research and drug design. In this study, by using the multi-label theory, we developed a new predictor called "pLoc-mGneg" for predicting the subcellular localization of Gram-negative bacterial proteins with both single and multiple locations. Rigorous cross-validation on a high quality benchmark dataset indicated that the proposed predictor is remarkably superior to "iLoc-Gneg", the state-of-the-art predictor for the same purpose. For the convenience of most experimental scientists, a user-friendly web-server for the novel predictor has been established at http://www.jci-bioinfo.cn/pLoc-mGneg/, by which users can easily get their desired results without the need to go through the complicated mathematics involved.

Accurate Classification of Protein Subcellular Localization from High-Throughput Microscopy Images Using Deep Learning

High-throughput microscopy of many single cells generates high-dimensional data that are far from straightforward to analyze. One important problem is automatically detecting the cellular compartment where a fluorescently-tagged protein resides, a task relatively simple for an experienced human, but difficult to automate on a computer. Here, we train an 11-layer neural network on data from mapping thousands of yeast proteins, achieving per cell localization classification accuracy of 91%, and per protein accuracy of 99% on held-out images. We confirm that low-level network features correspond to basic image characteristics, while deeper layers separate localization classes. Using this network as a feature calculator, we train standard classifiers that assign proteins to previously unseen compartments after observing only a small number of training examples. Our results are the most accurate subcellular localization classifications to date, and demonstrate the usefulness of deep learning for high-throughput microscopy.

Glucose-6-phosphate dehydrogenase as a target for highly efficient fatty acid biosynthesis in microalgae by enhancing NADPH supply

Oleaginous microalgae have great prospects in the fields of feed, nutrition, biofuel, etc. However, biomass and lipid productivity in microalgae remain a major economic and technological bottleneck. Here we present a novel regulatory target, glucose-6-phosphate dehydrogenase (G6PD) from the pentose phosphate pathway (PPP), in boosting microalgal lipid accumulation. G6PD, involved in the formation of NADPH demanded in fatty acid biosynthesis as reducing power, was characterized in oleaginous microalga Phaeodactylum tricornutum. In G6PD overexpressing microalgae, transcript abundance of G6PD increased by 4.4-fold, and G6PD enzyme activity increased by more than 3.1-fold with enhanced NADPH production. Consequently, the lipid content increased by 2.7-fold and reached up to 55.7% of dry weight, while cell growth was not apparently affected. The fatty acid composition exhibited significant changes, including a remarkable increase in monounsaturated fatty acids C16:1 and C18:1 concomitant with a decrease in polyunsaturated fatty acids C20:5 and C22:6. G6PD was localized to the chloroplast and its overexpression stimulated an increase in the number and size of oil bodies. Proteomic and metabolomic analyzes revealed that G6PD play a key role in regulating pentose phosphate pathway and subsequently upregulating NADPH consuming pathways such as fatty acid synthesis, thus eventually leading to lipid accumulation. Our findings show the critical role of G6PD in microalgal lipid accumulation by enhancing NADPH supply and demonstrate that G6PD is a promising target for metabolic engineering.

A novel multiphoton microscopy images segmentation method based on superpixel and watershed

Multiphoton microscopy (MPM) imaging technique based on two-photon excited fluorescence (TPEF) and second harmonic generation (SHG) shows fantastic performance for biological imaging. The automatic segmentation of cellular architectural properties for biomedical diagnosis based on MPM images is still a challenging issue. A novel multiphoton microscopy images segmentation method based on superpixels and watershed (MSW) is presented here to provide good segmentation results for MPM images. The proposed method uses SLIC superpixels instead of pixels to analyze MPM images for the first time. The superpixels segmentation based on a new distance metric combined with spatial, CIE Lab color space and phase congruency features, divides the images into patches which keep the details of the cell boundaries. Then the superpixels are used to reconstruct new images by defining an average value of superpixels as image pixels intensity level. Finally, the marker-controlled watershed is utilized to segment the cell boundaries from the reconstructed images. Experimental results show that cellular boundaries can be extracted from MPM images by MSW with higher accuracy and robustness.

A quantitative framework for the analysis of multimodal optical microscopy images

BACKGROUND

Multimodal optical microscopy, a set of imaging techniques based on unique, yet complementary contrast mechanisms and spatially and temporally co-registered data acquisition, has emerged as a powerful biomedical tool. However, the analysis of the dense, high-dimensional datasets acquired by these instruments remains mostly qualitative and restricted to analysis of each modality individually.

METHODS

Using a custom-built multimodal nonlinear optical microscope, high dimensional datasets were acquired for automated classification of functional cell states as well as identification of histopathological features in tissues slices. Supervised classification of cell death modes was performed through support vector machines (SVM) and semi-supervised classification of tissue slices was performed through the use of the expectation maximization (EM) algorithm.

RESULTS

Applications of these techniques to the automated classification of cell death modes as well as to the identification of tissue components in fixed ex vivo tissue slices are presented. The analysis techniques developed provide a direct link between multimodal image contrast and biological structure and function, resulting in highly accurate classification in both settings.

CONCLUSIONS

Quantification of multimodal optical microscopy images through statistical modeling of the high dimensional data acquired gives a strong correlation between biological structure and function and image contrast. These methods are sensitive to the identification of diagnostic, cellular-level features important in a variety of clinical settings.

pLoc-mPlant: predict subcellular localization of multi-location plant proteins by incorporating the optimal GO information into general PseAAC

One of the fundamental goals in cellular biochemistry is to identify the functions of proteins in the context of compartments that organize them in the cellular environment.

Strategy and software for the statistical spatial analysis of 3D intracellular distributions

The localization of proteins in specific domains or compartments in the 3D cellular space is essential for many fundamental processes in eukaryotic cells. Deciphering spatial organization principles within cells is a challenging task, in particular because of the large morphological variations between individual cells. We present here an approach for normalizing variations in cell morphology and for statistically analyzing spatial distributions of intracellular compartments from collections of 3D images. The method relies on the processing and analysis of 3D geometrical models that are generated from image stacks and that are used to build representations at progressively increasing levels of integration, ultimately revealing statistical significant traits of spatial distributions. To make this methodology widely available to end-users, we implemented our algorithmic pipeline into a user-friendly, multi-platform, and freely available software. To validate our approach, we generated 3D statistical maps of endomembrane compartments at subcellular resolution within an average epidermal root cell from collections of image stacks. This revealed unsuspected polar distribution patterns of organelles that were not detectable in individual images. By reversing the classical 'measure-then-average' paradigm, one major benefit of the proposed strategy is the production and display of statistical 3D representations of spatial organizations, thus fully preserving the spatial dimension of image data and at the same time allowing their integration over individual observations. The approach and software are generic and should be of general interest for experimental and modeling studies of spatial organizations at multiple scales (subcellular, cellular, tissular) in biological systems.

Computational spatiotemporal analysis identifies WAVE2 and cofilin as joint regulators of costimulation-mediated T cell actin dynamics

Fluorescence microscopy is one of the most important tools in cell biology research because it provides spatial and temporal information to investigate regulatory systems inside cells. This technique can generate data in the form of signal intensities at thousands of positions resolved inside individual live cells. However, given extensive cell-to-cell variation, these data cannot be readily assembled into three- or four-dimensional maps of protein concentration that can be compared across different cells and conditions. We have developed a method to enable comparison of imaging data from many cells and applied it to investigate actin dynamics in T cell activation. Antigen recognition in T cells by the T cell receptor (TCR) is amplified by engagement of the costimulatory receptor CD28. We imaged actin and eight core actin regulators to generate over a thousand movies of T cells under conditions in which CD28 was either engaged or blocked in the context of a strong TCR signal. Our computational analysis showed that the primary effect of costimulation blockade was to decrease recruitment of the activator of actin nucleation WAVE2 (Wiskott-Aldrich syndrome protein family verprolin-homologous protein 2) and the actin-severing protein cofilin to F-actin. Reconstitution of WAVE2 and cofilin activity restored the defect in actin signaling dynamics caused by costimulation blockade. Thus, we have developed and validated an approach to quantify protein distributions in time and space for the analysis of complex regulatory systems.

Building cell models and simulations from microscope images

The use of fluorescence microscopy has undergone a major revolution over the past twenty years, both with the development of dramatic new technologies and with the widespread adoption of image analysis and machine learning methods. Many open source software tools provide the ability to use these methods in a wide range of studies, and many molecular and cellular phenotypes can now be automatically distinguished. This article presents the next major challenge in microscopy automation, the creation of accurate models of cell organization directly from images, and reviews the progress that has been made towards this challenge.

Visualizing quantitative microscopy data: History and challenges

Data visualization is a fundamental aspect of science. In the context of microscopy-based studies, visualization typically involves presentation of the images themselves. However, data visualization is challenging when microscopy experiments entail imaging of millions of cells, and complex cellular phenotypes are quantified in a high-content manner. Most well-established visualization tools are inappropriate for displaying high-content data, which has driven the development of new visualization methodology. In this review, we discuss how data has been visualized in both classical and high-content microscopy studies; as well as the advantages, and disadvantages, of different visualization methods.

CP-CHARM: segmentation-free image classification made accessible

Background

Automated classification using machine learning often relies on features derived from segmenting individual objects, which can be difficult to automate. WND-CHARM is a previously developed classification algorithm in which features are computed on the whole image, thereby avoiding the need for segmentation. The algorithm obtained encouraging results but requires considerable computational expertise to execute. Furthermore, some benchmark sets have been shown to be subject to confounding artifacts that overestimate classification accuracy.

Results

We developed CP-CHARM, a user-friendly image-based classification algorithm inspired by WND-CHARM in (i) its ability to capture a wide variety of morphological aspects of the image, and (ii) the absence of requirement for segmentation. In order to make such an image-based classification method easily accessible to the biological research community, CP-CHARM relies on the widely-used open-source image analysis software CellProfiler for feature extraction. To validate our method, we reproduced WND-CHARM’s results and ensured that CP-CHARM obtained comparable performance. We then successfully applied our approach on cell-based assay data and on tissue images. We designed these new training and test sets to reduce the effect of batch-related artifacts.

Conclusions

The proposed method preserves the strengths of WND-CHARM - it extracts a wide variety of morphological features directly on whole images thereby avoiding the need for cell segmentation, but additionally, it makes the methods easily accessible for researchers without computational expertise by implementing them as a CellProfiler pipeline. It has been demonstrated to perform well on a wide range of bioimage classification problems, including on new datasets that have been carefully selected and annotated to minimize batch effects. This provides for the first time a realistic and reliable assessment of the whole image classification strategy.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-0895-y) contains supplementary material, which is available to authorized users.

Overexpression of Cotton GhMPK11 Decreases Disease Resistance through the Gibberellin Signaling Pathway in Transgenic Nicotiana benthamiana

Many changes in development, growth, hormone activity and environmental stimuli responses are mediated by mitogen-activated protein kinase (MAPK) cascades. However, in plants, studies on MAPKs have mainly focused on MPK3, MPK4 and MPK6. Here, a novel group B MAPK gene, GhMPK11, was isolated from cotton (Gossypium hirsutum L.) and characterized. Both promoter and expression pattern analyses revealed that GhMPK11 is involved in defense responses and signaling pathways. GhMPK11 overexpression in Nicotiana benthamiana plants could increase gibberellin 3 (GA3) content through the regulation of GA-related genes. Interestingly, either GhMPK11 overexpression or exogenous GA3 treatment in N. benthamiana plants could enhance the susceptibility of these plants to the infectious pathogens Ralstonia solanacearum and Rhizoctonia solani. Moreover, reactive oxygen species (ROS) accumulation was increased after pathogen infiltration due to the increased expression of ROS-related gene respiratory burst oxidative homologs (RbohB) and the decreased expression or activity of ROS detoxification enzymes regulated by GA3, such as superoxide dismutases (SODs), peroxidases (PODs), catalase (CAT) and glutathione S-transferase (GST). Taken together, these results suggest that GhMPK11 overexpression could enhance the susceptibility of tobacco to pathogen infection through the GA3 signaling pathway via down-regulation of ROS detoxification enzymes.

Time series modeling of live-cell shape dynamics for image-based phenotypic profiling

Live-cell imaging can be used to capture spatio-temporal aspects of cellular responses that are not accessible to fixed-cell imaging. As the use of live-cell imaging continues to increase, new computational procedures are needed to characterize and classify the temporal dynamics of individual cells. For this purpose, here we present the general experimental-computational framework SAPHIRE (Stochastic Annotation of Phenotypic Individual-cell Responses) to characterize phenotypic cellular responses from time series imaging datasets. Hidden Markov modeling is used to infer and annotate morphological state and state-switching properties from image-derived cell shape measurements. Time series modeling is performed on each cell individually, making the approach broadly useful for analyzing asynchronous cell populations. Two-color fluorescent cells simultaneously expressing actin and nuclear reporters enabled us to profile temporal changes in cell shape following pharmacological inhibition of cytoskeleton-regulatory signaling pathways. Results are compared with existing approaches conventionally applied to fixed-cell imaging datasets, and indicate that time series modeling captures heterogeneous dynamic cellular responses that can improve drug classification and offer additional important insight into mechanisms of drug action. The software is available at http://saphire-hcs.org.

A generic methodological framework for studying single cell motility in high-throughput time-lapse data

Motivation: Motility is a fundamental cellular attribute, which plays a major part in processes ranging from embryonic development to metastasis. Traditionally, single cell motility is often studied by live cell imaging. Yet, such studies were so far limited to low throughput. To systematically study cell motility at a large scale, we need robust methods to quantify cell trajectories in live cell imaging data.

Results: The primary contribution of this article is to present Motility study Integrated Workflow (MotIW), a generic workflow for the study of single cell motility in high-throughput time-lapse screening data. It is composed of cell tracking, cell trajectory mapping to an original feature space and hit detection according to a new statistical procedure. We show that this workflow is scalable and demonstrates its power by application to simulated data, as well as large-scale live cell imaging data. This application enables the identification of an ontology of cell motility patterns in a fully unsupervised manner.

Availability and implementation: Python code and examples are available online (http://cbio.ensmp.fr/∼aschoenauer/motiw.html)

Contact:thomas.walter@mines-paristech.fr

Supplementary information:Supplementary data are available at Bioinformatics online.

Human cell structure-driven model construction for predicting protein subcellular location from biological images

MOTIVATION

The systematic study of subcellular location pattern is very important for fully characterizing the human proteome. Nowadays, with the great advances in automated microscopic imaging, accurate bioimage-based classification methods to predict protein subcellular locations are highly desired. All existing models were constructed on the independent parallel hypothesis, where the cellular component classes are positioned independently in a multi-class classification engine. The important structural information of cellular compartments is missed. To deal with this problem for developing more accurate models, we proposed a novel cell structure-driven classifier construction approach (SC-PSorter) by employing the prior biological structural information in the learning model. Specifically, the structural relationship among the cellular components is reflected by a new codeword matrix under the error correcting output coding framework. Then, we construct multiple SC-PSorter-based classifiers corresponding to the columns of the error correcting output coding codeword matrix using a multi-kernel support vector machine classification approach. Finally, we perform the classifier ensemble by combining those multiple SC-PSorter-based classifiers via majority voting.

RESULTS

We evaluate our method on a collection of 1636 immunohistochemistry images from the Human Protein Atlas database. The experimental results show that our method achieves an overall accuracy of 89.0%, which is 6.4% higher than the state-of-the-art method.

AVAILABILITY AND IMPLEMENTATION

The dataset and code can be downloaded from https://github.com/shaoweinuaa/.

CONTACT

dqzhang@nuaa.edu.cn

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Performance and sensitivity evaluation of 3D spot detection methods in confocal microscopy

Reliable 3D detection of diffraction-limited spots in fluorescence microscopy images is an important task in subcellular observation. Generally, fluorescence microscopy images are heavily degraded by noise and non-specifically stained background, making reliable detection a challenging task. In this work, we have studied the performance and parameter sensitivity of eight recent methods for 3D spot detection. The study is based on both 3D synthetic image data and 3D real confocal microscopy images. The synthetic images were generated using a simulator modeling the complete imaging setup, including the optical path as well as the image acquisition process. We studied the detection performance and parameter sensitivity under different noise levels and under the influence of uneven background signal. To evaluate the parameter sensitivity, we propose a novel measure based on the gradient magnitude of the F1 score. We measured the success rate of the individual methods for different types of the image data and found that the type of image degradation is an important factor. Using the F1 score and the newly proposed sensitivity measure, we found that the parameter sensitivity is not necessarily proportional to the success rate of a method. This also provided an explanation why the best performing method for synthetic data was outperformed by other methods when applied to the real microscopy images. On the basis of the results obtained, we conclude with the recommendation of the HDome method for data with relatively low variations in quality, or the Sorokin method for image sets in which the quality varies more. We also provide alternative recommendations for high-quality images, and for situations in which detailed parameter tuning might be deemed expensive.

Rapid 3-D delineation of cell nuclei for high-content screening platforms

High-resolution three-dimensional (3-D) microscopy combined with multiplexing of fluorescent labels allows high-content analysis of large numbers of cell nuclei. The full automation of 3-D screening platforms necessitates image processing algorithms that can accurately and robustly delineate nuclei in images with little to no human intervention. Imaging-based high-content screening was originally developed as a powerful tool for drug discovery. However, cell confluency, complexity of nuclear staining as well as poor contrast between nuclei and background result in slow and unreliable 3-D image processing and therefore negatively affect the performance of studying a drug response. Here, we propose a new method, 3D-RSD, to delineate nuclei by means of 3-D radial symmetries and test it on high-resolution image data of human cancer cells treated by drugs. The nuclei detection performance was evaluated by means of manually generated ground truth from 2351 nuclei (27 confocal stacks). When compared to three other nuclei segmentation methods, 3D-RSD possessed a better true positive rate of 83.3% and F-score of 0.895±0.045 (p-value=0.047). Altogether, 3D-RSD is a method with a very good overall segmentation performance. Furthermore, implementation of radial symmetries offers good processing speed, and makes 3D-RSD less sensitive to staining patterns. In particular, the 3D-RSD method performs well in cell lines, which are often used in imaging-based HCS platforms and are afflicted by nuclear crowding and overlaps that hinder feature extraction.

Cell-based screening: extracting meaning from complex data

Unbiased discovery approaches have the potential to uncover neurobiological insights into CNS disease and lead to the development of therapies. Here, we review lessons learned from imaging-based screening approaches and recent advances in these areas, including powerful new computational tools to synthesize complex data into more useful knowledge that can reliably guide future research and development.

EUROPattern Suite technology for computer-aided immunofluorescence microscopy in autoantibody diagnostics

Antinuclear autoantibodies (ANA) are highly informative biomarkers in autoimmune diagnostics. The increasing demand for effective test systems, however, has led to the development of a confusingly large variety of different platforms. One of them, the indirect immunofluorescence (IIF), is regarded as the common gold standard for ANA screening, as described in a position statement by the American College of Rheumatology in 2009. Technological solutions have been developed aimed at standardization and automation of IIF to overcome methodological limitations and subjective bias in IIF interpretation. In this review, we present the EUROPattern Suite, a system for computer-aided immunofluorescence microscopy (CAIFM) including automated acquisition of digital images and evaluation of IIF results. The system was originally designed for ANA diagnostics on human epithelial cells, but its applications have been extended with the latest system update version 1.5 to the analysis of antineutrophil cytoplasmic antibodies (ANCA) and anti-dsDNA antibodies.

Automatic determination of NET (neutrophil extracellular traps) coverage in fluorescent microscopy images

MOTIVATION

Neutrophil extracellular traps (NETs) are believed to be essential in controlling several bacterial pathogens. Quantification of NETs in vitro is an important tool in studies aiming to clarify the biological and chemical factors contributing to NET production, stabilization and degradation. This estimation can be performed on the basis of fluorescent microscopy images using appropriate labelings. In this context, it is desirable to automate the analysis to eliminate both the tedious process of manual annotation and possible operator-specific biases.

RESULTS

We propose a framework for the automated determination of NET content, based on visually annotated images which are used to train a supervised machine-learning method. We derive several methods in this framework. The best results are obtained by combining these into a single prediction. The overall Q(2) of the combined method is 93%. By having two experts label part of the image set, we were able to compare the performance of the algorithms to the human interoperator variability. We find that the two operators exhibited a very high correlation on their overall assessment of the NET coverage area in the images (R(2) is 97%), although there were consistent differences in labeling at pixel level (Q(2), which unlike R(2) does not correct for additive and multiplicative biases, was only 89%).

AVAILABILITY AND IMPLEMENTATION

Open source software (under the MIT license) is available at https://github.com/luispedro/Coelho2015_NetsDetermination for both reproducibility and application to new data.

Visualizing cellular imaging data using PhenoPlot

Visualization is essential for data interpretation, hypothesis formulation and communication of results. However, there is a paucity of visualization methods for image-derived data sets generated by high-content analysis in which complex cellular phenotypes are described as high-dimensional vectors of features. Here we present a visualization tool, PhenoPlot, which represents quantitative high-content imaging data as easily interpretable glyphs, and we illustrate how PhenoPlot can be used to improve the exploration and interpretation of complex breast cancer cell phenotypes.

A new multi-label classifier in identifying the functional types of human membrane proteins

Membrane proteins were found to be involved in various cellular processes performing various important functions, which are mainly associated to their type. Given a membrane protein sequence, how can we identify its type(s)? Particularly, how can we deal with the multi-type problem since one membrane protein may simultaneously belong to two or more different types? To address these problems, which are obviously very important to both basic research and drug development, a new multi-label classifier was developed based on pseudo amino acid composition with multi-label k-nearest neighbor algorithm. The success rate achieved by the new predictor on the benchmark dataset by jackknife test is 73.94%, indicating that the method is promising and the predictor may become a very useful high-throughput tool, or at least play a complementary role to the existing predictors in identifying functional types of membrane proteins.

On comparing heterogeneity across biomarkers

Microscopy reveals complex patterns of cellular heterogeneity that can be biologically informative. However, a limitation of microscopy is that only a small number of biomarkers can typically be monitored simultaneously. Thus, a natural question is whether additional biomarkers provide a deeper characterization of the distribution of cellular states in a population. How much information about a cell's phenotypic state in one biomarker is gained by knowing its state in another biomarker? Here, we describe a framework for comparing phenotypic states across biomarkers. Our approach overcomes the current limitation of microscopy by not requiring costaining biomarkers on the same cells; instead, we require staining of biomarkers (possibly separately) on a common collection of phenotypically diverse cell lines. We evaluate our approach on two image datasets: 33 oncogenically diverse lung cancer cell lines stained with 7 biomarkers, and 49 less diverse subclones of one lung cancer cell line stained with 12 biomarkers. We first validate our method by comparing it to the "gold standard" of costaining. We then apply our approach to all pairs of biomarkers and use it to identify biomarkers that yield similar patterns of heterogeneity. The results presented in this work suggest that many biomarkers provide redundant information about heterogeneity. Thus, our approach provides a practical guide for selecting independently informative biomarkers and, more generally, will yield insights into both the connectivity of biological networks and the complexity of the state space of biological systems.

Bioimaging-based detection of mislocalized proteins in human cancers by semi-supervised learning

MOTIVATION

There is a long-term interest in the challenging task of finding translocated and mislocated cancer biomarker proteins. Bioimages of subcellular protein distribution are new data sources which have attracted much attention in recent years because of their intuitive and detailed descriptions of protein distribution. However, automated methods in large-scale biomarker screening suffer significantly from the lack of subcellular location annotations for bioimages from cancer tissues. The transfer prediction idea of applying models trained on normal tissue proteins to predict the subcellular locations of cancerous ones is arbitrary because the protein distribution patterns may differ in normal and cancerous states.

RESULTS

We developed a new semi-supervised protocol that can use unlabeled cancer protein data in model construction by an iterative and incremental training strategy. Our approach enables us to selectively use the low-quality images in normal states to expand the training sample space and provides a general way for dealing with the small size of annotated images used together with large unannotated ones. Experiments demonstrate that the new semi-supervised protocol can result in improved accuracy and sensitivity of subcellular location difference detection.

AVAILABILITY AND IMPLEMENTATION

The data and code are available at: www.csbio.sjtu.edu.cn/bioinf/SemiBiomarker/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Genetic improvement of the microalga Phaeodactylum tricornutum for boosting neutral lipid accumulation

To obtain fast growing oil-rich microalgal strains has been urgently demanded for microalgal biofuel. Malic enzyme (ME), which is involved in pyruvate metabolism and carbon fixation, was first characterized in microalgae here. Overexpression of Phaeodactylum tricornutum ME (PtME) significantly enhanced the expression of PtME and its enzymatic activity in transgenic P. tricornutum. The total lipid content in transgenic cells markedly increased by 2.5-fold and reached a record 57.8% of dry cell weight with a similar growth rate to wild type, thus keeping a high biomass. The neutral lipid content was further increased by 31% under nitrogen-deprivation treatment, still 66% higher than that of wild type. Transgenic microalgae cells exhibited obvious morphological changes, as the cells were shorter and thicker and contained larger oil bodies. Immuno-electron microscopy targeted PtME to the mitochondrion. This study markedly increased the oil content in microalgae, suggesting a new route for developing ideal microalgal strains for industrial biodiesel production.

Predict subcellular locations of singleplex and multiplex proteins by semi-supervised learning and dimension-reducing general mode of Chou's PseAAC

Predicting protein subcellular location is one of major challenges in Bioinformatics area since such knowledge helps us understand protein functions and enables us to select the targeted proteins during drug discovery process. While many computational techniques have been proposed to improve predictive performance for protein subcellular location, they have several shortcomings. In this work, we propose a method to solve three main issues in such techniques; i) manipulation of multiplex proteins which may exist or move between multiple cellular compartments, ii) handling of high dimensionality in input and output spaces and iii) requirement of sufficient labeled data for model training. Towards these issues, this work presents a new computational method for predicting proteins which have either single or multiple locations. The proposed technique, namely iFLAST-CORE, incorporates the dimensionality reduction in the feature and label spaces with co-training paradigm for semi-supervised multi-label classification. For this purpose, the Singular Value Decomposition (SVD) is applied to transform the high-dimensional feature space and label space into the lower-dimensional spaces. After that, due to limitation of labeled data, the co-training regression makes use of unlabeled data by predicting the target values in the lower-dimensional spaces of unlabeled data. In the last step, the component of SVD is used to project labels in the lower-dimensional space back to those in the original space and an adaptive threshold is used to map a numeric value to a binary value for label determination. A set of experiments on viral proteins and gram-negative bacterial proteins evidence that our proposed method improve the classification performance in terms of various evaluation metrics such as Aiming (or Precision), Coverage (or Recall) and macro F-measure, compared to the traditional method that uses only labeled data.

Delta 5 fatty acid desaturase upregulates the synthesis of polyunsaturated fatty acids in the marine diatom Phaeodactylum tricornutum

Microalgae are important primary producers in the marine ecosystem and excellent sources of lipids and other bioactive compounds. The marine diatom Phaeodactylum tricornutum accumulates eicosapentaenoic acid (EPA, 20:5n-3) as its major component of fatty acids. To improve the EPA production, delta 5 desaturase, which plays a role in EPA biosynthetic pathway, was characterized in P. tricornutum. An annotated delta 5 desaturase PtD5b gene was cloned and overexpressed in P. tricornutum. The transgene was integrated into the genome demonstrated by Southern blot, and the overexpression of PtD5b was verified by qPCR and Western blot analysis. Fatty acid composition exhibited a significant increase in the unsaturated fatty acids. Monounsaturated fatty acids (MUFA) and polyunsaturated fatty acids (PUFA) showed an increase of 75% and 64%, respectively. In particular, EPA showed an increase of 58% in engineered microalgae. Meanwhile, neutral lipid content showed an increase up to 65% in engineered microalgae. More importantly, engineered cells showed a similar growth rate with the wild type, thus keeping high biomass productivity. This work provides an effective way to improve the production of microalgal value-added bioproducts by metabolic engineering.

Many local pattern texture features: which is better for image-based multilabel human protein subcellular localization classification?

Human protein subcellular location prediction can provide critical knowledge for understanding a protein's function. Since significant progress has been made on digital microscopy, automated image-based protein subcellular location classification is urgently needed. In this paper, we aim to investigate more representative image features that can be effectively used for dealing with the multilabel subcellular image samples. We prepared a large multilabel immunohistochemistry (IHC) image benchmark from the Human Protein Atlas database and tested the performance of different local texture features, including completed local binary pattern, local tetra pattern, and the standard local binary pattern feature. According to our experimental results from binary relevance multilabel machine learning models, the completed local binary pattern, and local tetra pattern are more discriminative for describing IHC images when compared to the traditional local binary pattern descriptor. The combination of these two novel local pattern features and the conventional global texture features is also studied. The enhanced performance of final binary relevance classification model trained on the combined feature space demonstrates that different features are complementary to each other and thus capable of improving the accuracy of classification.

The actin-driven spatiotemporal organization of T-cell signaling at the system scale

T cells are activated through interaction with antigen-presenting cells (APCs). During activation, receptors and signaling intermediates accumulate in diverse spatiotemporal distributions. These distributions control the probability of signaling interactions and thus govern information flow through the signaling system. Spatiotemporally resolved system-scale investigation of signaling can extract the regulatory information thus encoded, allowing unique insight into the control of T-cell function. Substantial technical challenges exist, and these are briefly discussed herein. While much of the work assessing T-cell spatiotemporal organization uses planar APC substitutes, we focus here on B-cell APCs with often stark differences. Spatiotemporal signaling distributions are driven by cell biologically distinct structures, a large protein assembly at the interface center, a large invagination, the actin-supported interface periphery as extended by smaller individual lamella, and a newly discovered whole-interface actin-driven lamellum. The more than 60 elements of T-cell activation studied to date are dynamically distributed between these structures, generating a complex organization of the signaling system. Signal initiation and core signaling prefer the interface center, while signal amplification is localized in the transient lamellum. Actin dynamics control signaling distributions through regulation of the underlying structures and drive a highly undulating T-cell/APC interface that imposes substantial constraints on T-cell organization. We suggest that the regulation of actin dynamics, by controlling signaling distributions and membrane topology, is an important rheostat of T-cell signaling.

Current state of diagnostic technologies in the autoimmunology laboratory

The methods for detecting and measuring autoantibodies have evolved markedly in recent years, encompassing three generations of analytical technologies. Many different immunoassay methods have been developed and used for research and laboratory practice purposes, from the early conventional (or monoplex) analytical methods able to detect single autoantibodies to the more recent multiplex platforms that can quantify tens of molecules. Although it has been in use for over 50 years, indirect immunofluorescence remains the standard method for research on many types of autoantibodies, due to its characteristics of diagnostic sensitivity and also to recent technological innovations which permit it a greater level of automation and standardization. The recent multiplex immunometric methods, with varying levels of automation, present characteristics of higher diagnostic accuracy, but are not yet widely diffused in autoimmunology laboratories due to the limited number of autoantibodies that are detectable, and due to the high cost of reagents and systems. Technological advancement in autoimmunology continues to evolve rapidly, and in the coming years new proteomic techniques will be able to radically change the approach to diagnostics and possibly also clinical treatment of autoimmune diseases. The scope of this review is to update the state of the art of technologies and methods for the measurement of autoantibodies, with special reference to innovations in indirect immunofluorescence and in multiple proteomic methods.

Single cell analysis with probe ESI-mass spectrometry: detection of metabolites at cellular and subcellular levels

Molecular analysis at cellular and subcellular levels, whether on selected molecules or at the metabolomics scale, is still a challenge now. Here we propose a method based on probe ESI mass spectrometry (PESI-MS) for single cell analysis. Detection of metabolites at cellular and subcellular levels was successfully achieved. In our work, tungsten probes with a tip diameter of about 1 μm were directly inserted into live cells to enrich metabolites. Then the enriched metabolites were directly desorbed/ionized from the tip of the probe for mass spectrometry (MS) detection. The direct desorption/ionization of the enriched metabolites from the tip of the probe greatly improved the sensitivity by a factor of about 30 fold compared to those methods that eluted the enriched analytes into a liquid phase for subsequent MS detection. We applied the PESI-MS to the detection of metabolites in single Allium cepa cells. Different kinds of metabolites, including 6 fructans, 4 lipids, and 8 flavone derivatives in single cells, have been successfully detected. Significant metabolite diversity was observed among different cells types of A. cepa bulb and different subcellular compartments of the same cell. We found that the inner epidermal cells had about 20 fold more fructans than the outer epidermal cells, while the outer epidermal cells had more lipids. We expected that PESI-MS might be a candidate in the future studies of single cell "omics".

Multiparametric Analysis of Screening Data: Growing Beyond the Single Dimension to Infinity and Beyond

Advances in instrumentation now allow the development of screening assays that are capable of monitoring multiple readouts such as transcript or protein levels, or even multiple parameters derived from images. Such advances in assay technologies highlight the complex nature of biology and disease. Harnessing this complexity requires integration of all the different parameters that can be measured rather than just monitoring a single dimension as is commonly used. Although some of the methods used to combine multiple measurements, such as principal component analysis, are commonly used for microarray analysis, biologists are not yet using many of the tools that have been developed in other fields to address such issues. Visualization of multiparametric data sets is one of the major challenges in this field, and a depiction of the results in a manner that can be readily interpreted is essential. This article describes a number of assay systems being used to generate such data sets en masse, and the methods being applied to their visualization and analysis. We also discuss some of the challenges of applying methods developed in other fields to biology.

PPI network analysis of mRNA expression profile of ezrin knockdown in esophageal squamous cell carcinoma

Ezrin, coding protein EZR which cross-links actin filaments, overexpresses and involves invasion, metastasis, and poor prognosis in various cancers including esophageal squamous cell carcinoma (ESCC). In our previous study, Ezrin was knock down and analyzed by mRNA expression profile which has not been fully mined. In this study, we applied protein-protein interactions (PPI) network knowledge and methods to explore our understanding of these differentially expressed genes (DEGs). PPI subnetworks showed that hundreds of DEGs interact with thousands of other proteins. Subcellular localization analyses found that the DEGs and their directly or indirectly interacting proteins distribute in multiple layers, which was applied to analyze the shortest paths between EZR and other DEGs. Gene ontology annotation generated a functional annotation map and found hundreds of significant terms, especially those associated with cytoskeleton organization of Ezrin protein, such as "cytoskeleton organization," "regulation of actin filament-based process," and "regulation of actin cytoskeleton organization." The algorithm of Random Walk with Restart was applied to prioritize the DEGs and identified several cancer related DEGs ranked closest to EZR. These analyses based on PPI network have greatly expanded our comprehension of the mRNA expression profile of Ezrin knockdown for future examination of the roles and mechanisms of Ezrin.

Comparative genomics study for identification of drug and vaccine targets in Vibrio cholerae: MurA ligase as a case study

A systematic workflow consisting of comparative genomics, metabolic pathways analysis and additional drug prioritization parameters identified 264 proteins of Vibrio cholerae which were predicted to be absent in Homo sapiens. Among these, 40 proteins were identified as essential proteins that could serve as potential drug and vaccine targets. Additional prioritization parameters characterized 11 proteins as vaccine candidates while druggability of each of the identified proteins as evaluated by the Drug Bank database which prioritized 16 proteins suitable for drug targets. As a case study, we built a homology model of one of the potential drug targets, MurA ligase, using MODELLER (9v12) software. The model has been further explored for in silico docking with inhibitors having druggability potential from the Drug Bank database. Results from this study could facilitate selecting V. cholerae proteins for drug design and vaccine production pipelines in future.