DeepHistoClass: A Novel Strategy for Confident Classification of Immunohistochemistry Images Using Deep Learning.

A multitude of efforts worldwide aim to create a single-cell reference map of the human body, for fundamental understanding of human health, molecular medicine, and targeted treatment. Antibody-based proteomics using immunohistochemistry (IHC) has proven to be an excellent technology for integration with large-scale single-cell transcriptomics datasets. The golden standard for evaluation of IHC staining patterns is manual annotation, which is expensive and may lead to subjective errors. Artificial intelligence holds much promise for efficient and accurate pattern recognition, but confidence in prediction needs to be addressed. Here, the aim was to present a reliable and comprehensive framework for automated annotation of IHC images. We developed a multilabel classification of 7848 complex IHC images of human testis corresponding to 2794 unique proteins, generated as part of the Human Protein Atlas (HPA) project. Manual annotation data for eight different cell types was generated as a basis for training and testing a proposed Hybrid Bayesian Neural Network. By combining the deep learning model with a novel uncertainty metric, DeepHistoClass (DHC) Confidence Score, the average diagnostic performance improved from 86.9% to 96.3%. This metric not only reveals which images are reliably classified by the model, but can also be utilized for identification of manual annotation errors. The proposed streamlined workflow can be developed further for other tissue types in health and disease and has important implications for digital pathology initiatives or large-scale protein mapping efforts such as the HPA project.

Identification of cancer biomarkers of prognostic value using specific gene regulatory networks (GRN): a novel role of RAD51AP1 for ovarian and lung cancers.

To date, microarray analyses have led to the discovery of numerous individual 'molecular signatures' associated with specific cancers. However, there are serious limitations for the adoption of these multi-gene signatures in the clinical environment for diagnostic or prognostic testing as studies with more power need to be carried out. This may involve larger richer cohorts and more advanced analyses. In this study, we conduct analyses-based on gene regulatory network-to reveal distinct and common biomarkers across cancer types. Using microarray data of triple-negative and medullary breast, ovarian and lung cancers applied to a combination of glasso and Bayesian networks (BNs), we derived a unique network-containing genes that are uniquely involved: small proline-rich protein 1A (SPRR1A), follistatin like 1 (FSTL1), collagen type XII alpha 1 (COL12A1) and RAD51 associated protein 1 (RAD51AP1). RAD51AP1 and FSTL1 are significantly overexpressed in ovarian cancer patients but only RAD51AP1 is upregulated in lung cancer patients compared with healthy controls. The upregulation of RAD51AP1 was mirrored in the bloods of both ovarian and lung cancer patients, and Kaplan-Meier (KM) plots predicted poorer overall survival (OS) in patients with high expression of RAD51AP1. Suppression of RAD51AP1 by RNA interference reduced cell proliferation in vitro in ovarian (SKOV3) and lung (A549) cancer cells. This effect appears to be modulated by a decrease in the expression of mTOR-related genes and pro-metastatic candidate genes. Our data describe how an initial in silico approach can generate novel biomarkers that could potentially support current clinical practice and improve long-term outcomes.

Impact of e-publication changes in the International Code of Nomenclature for algae, fungi and plants (Melbourne Code, 2012) - did we need to "run for our lives"?

BACKGROUND: At the Nomenclature Section of the XVIII International Botanical Congress in Melbourne, Australia (IBC), the botanical community voted to allow electronic publication of nomenclatural acts for algae, fungi and plants, and to abolish the rule requiring Latin descriptions or diagnoses for new taxa. Since the 1st January 2012, botanists have been able to publish new names in electronic journals and may use Latin or English as the language of description or diagnosis. RESULTS: Using data on vascular plants from the International Plant Names Index (IPNI) spanning the time period in which these changes occurred, we analysed trajectories in publication trends and assessed the impact of these new rules for descriptions of new species and nomenclatural acts. The data show that the ability to publish electronically has not "opened the floodgates" to an avalanche of sloppy nomenclature, but concomitantly neither has there been a massive expansion in the number of names published, nor of new authors and titles participating in publication of botanical nomenclature. CONCLUSIONS: The e-publication changes introduced in the Melbourne Code have gained acceptance, and botanists are using these new techniques to describe and publish their work. They have not, however, accelerated the rate of plant species description or participation in biodiversity discovery as was hoped.

Identifying and handling data bias within primary healthcare data using synthetic data generators.

Advanced synthetic data generators can simulate data samples that closely resemble sensitive personal datasets while significantly reducing the risk of individual identification. The use of these advanced generators holds enormous potential in the medical field, as it allows for the simulation and sharing of sensitive patient data. This enables the development and rigorous validation of novel AI technologies for accurate diagnosis and efficient disease management. Despite the availability of massive ground truth datasets (such as UK-NHS databases that contain millions of patient records), the risk of biases being carried over to data generators still exists. These biases may arise from the under-representation of specific patient cohorts due to cultural sensitivities within certain communities or standardised data collection procedures. Machine learning models can exhibit bias in various forms, including the under-representation of certain groups in the data. This can lead to missing data and inaccurate correlations and distributions, which may also be reflected in synthetic data. Our paper aims to improve synthetic data generators by introducing probabilistic approaches to first detect difficult-to-predict data samples in ground truth data and then boost them when applying the generator. In addition, we explore strategies to generate synthetic data that can reduce bias and, at the same time, improve the performance of predictive models.

Modelling and analysing the dynamics of disease progression from cross-sectional studies.

Clinical trials are typically conducted over a population within a defined time period in order to illuminate certain characteristics of a health issue or disease process. These cross-sectional studies give us a 'snapshot' of this disease process over a large number of people but do not allow us to model the temporal nature of disease, thereby allowing for modelling detailed prognostic predictions. The aim of this paper is to explore an extension of the temporal bootstrap to identify intermediate stages in a disease process and sub-categories of the disease exhibiting subtly different symptoms. Our approach is compared to a strawman method and investigated in its ability to explain the dynamics of progression on biomedical data from three diseases: Glaucoma, Breast Cancer and Parkinson's disease. We focus on creating reliable time-series models from large amounts of historical cross-sectional data using the temporal bootstrap technique. Two issues are explored: how to build time-series models from cross-sectional data, and how to automatically identify different disease states along these trajectories, as well as the transitions between them. Our approach of relabeling trajectories allows us to explore the temporal nature of how diseases progress even when time-series data is not available (if the cross-sectional study is large enough). We intend to expand this research to deal with multiple studies where we can combine both cross-sectional and longitudinal datasets and to focus on the junctions of the trajectories as key stages in the progression of disease.

Interspecies translation of disease networks increases robustness and predictive accuracy.

Gene regulatory networks give important insights into the mechanisms underlying physiology and pathophysiology. The derivation of gene regulatory networks from high-throughput expression data via machine learning strategies is problematic as the reliability of these models is often compromised by limited and highly variable samples, heterogeneity in transcript isoforms, noise, and other artifacts. Here, we develop a novel algorithm, dubbed Dandelion, in which we construct and train intraspecies Bayesian networks that are translated and assessed on independent test sets from other species in a reiterative procedure. The interspecies disease networks are subjected to multi-layers of analysis and evaluation, leading to the identification of the most consistent relationships within the network structure. In this study, we demonstrate the performance of our algorithms on datasets from animal models of oculopharyngeal muscular dystrophy (OPMD) and patient materials. We show that the interspecies network of genes coding for the proteasome provide highly accurate predictions on gene expression levels and disease phenotype. Moreover, the cross-species translation increases the stability and robustness of these networks. Unlike existing modeling approaches, our algorithms do not require assumptions on notoriously difficult one-to-one mapping of protein orthologues or alternative transcripts and can deal with missing data. We show that the identified key components of the OPMD disease network can be confirmed in an unseen and independent disease model. This study presents a state-of-the-art strategy in constructing interspecies disease networks that provide crucial information on regulatory relationships among genes, leading to better understanding of the disease molecular mechanisms.

The role of the North Atlantic Oscillation in controlling U.K. butterfly population size and phenology.

1. The North Atlantic Oscillation (NAO) exerts considerable control on U.K. weather. This study investigates the impact of the NAO on butterfly abundance and phenology using 34 years of data from the U.K. Butterfly Monitoring Scheme (UKBMS).2. The study uses a multi-species indicator to show that the NAO does not affect overall U.K. butterfly population size. However, the abundance of bivoltine butterfly species, which have longer flight seasons, were found to be more likely to respond positively to the NAO compared with univoltine species, which show little or a negative response.3. A positive winter NAO index is associated with warmer weather and earlier flight dates for Anthocharis cardamines (Lepidoptera: Pieridae), Melanargia galathea (Lepidoptera: Nymphalidae), Aphantopus hyperantus (Lepidoptera: Nymphalidae), Pyronia tithonus (Lepidoptera: Nymphalidae), Lasiommata megera (Lepidoptera: Nymphalidae) and Polyommatus icarus (Lepidoptera: Lycaenidae). In bivoltine species, the NAO affects the phenology of the first generation, the timing of which indirectly controls the timing of the second generation.4. The NAO influences the timing of U.K. butterfly flight seasons more strongly than it influences population size.

Harnessing Large-Scale Herbarium Image Datasets Through Representation Learning.

The mobilization of large-scale datasets of specimen images and metadata through herbarium digitization provide a rich environment for the application and development of machine learning techniques. However, limited access to computational resources and uneven progress in digitization, especially for small herbaria, still present barriers to the wide adoption of these new technologies. Using deep learning to extract representations of herbarium specimens useful for a wide variety of applications, so-called "representation learning," could help remove these barriers. Despite its recent popularity for camera trap and natural world images, representation learning is not yet as popular for herbarium specimen images. We investigated the potential of representation learning with specimen images by building three neural networks using a publicly available dataset of over 2 million specimen images spanning multiple continents and institutions. We compared the extracted representations and tested their performance in application tasks relevant to research carried out with herbarium specimens. We found a triplet network, a type of neural network that learns distances between images, produced representations that transferred the best across all applications investigated. Our results demonstrate that it is possible to learn representations of specimen images useful in different applications, and we identify some further steps that we believe are necessary for representation learning to harness the rich information held in the worlds' herbaria.

The identification of informative genes from multiple datasets with increasing complexity.

BACKGROUND: In microarray data analysis, factors such as data quality, biological variation, and the increasingly multi-layered nature of more complex biological systems complicates the modelling of regulatory networks that can represent and capture the interactions among genes. We believe that the use of multiple datasets derived from related biological systems leads to more robust models. Therefore, we developed a novel framework for modelling regulatory networks that involves training and evaluation on independent datasets. Our approach includes the following steps: (1) ordering the datasets based on their level of noise and informativeness; (2) selection of a Bayesian classifier with an appropriate level of complexity by evaluation of predictive performance on independent data sets; (3) comparing the different gene selections and the influence of increasing the model complexity; (4) functional analysis of the informative genes. RESULTS: In this paper, we identify the most appropriate model complexity using cross-validation and independent test set validation for predicting gene expression in three published datasets related to myogenesis and muscle differentiation. Furthermore, we demonstrate that models trained on simpler datasets can be used to identify interactions among genes and select the most informative. We also show that these models can explain the myogenesis-related genes (genes of interest) significantly better than others (P < 0.004) since the improvement in their rankings is much more pronounced. Finally, after further evaluating our results on synthetic datasets, we show that our approach outperforms a concordance method by Lai et al. in identifying informative genes from multiple datasets with increasing complexity whilst additionally modelling the interaction between genes. CONCLUSIONS: We show that Bayesian networks derived from simpler controlled systems have better performance than those trained on datasets from more complex biological systems. Further, we present that highly predictive and consistent genes, from the pool of differentially expressed genes, across independent datasets are more likely to be fundamentally involved in the biological process under study. We conclude that networks trained on simpler controlled systems, such as in vitro experiments, can be used to model and capture interactions among genes in more complex datasets, such as in vivo experiments, where these interactions would otherwise be concealed by a multitude of other ongoing events.

Generating high-fidelity synthetic patient data for assessing machine learning healthcare software.

There is a growing demand for the uptake of modern artificial intelligence technologies within healthcare systems. Many of these technologies exploit historical patient health data to build powerful predictive models that can be used to improve diagnosis and understanding of disease. However, there are many issues concerning patient privacy that need to be accounted for in order to enable this data to be better harnessed by all sectors. One approach that could offer a method of circumventing privacy issues is the creation of realistic synthetic data sets that capture as many of the complexities of the original data set (distributions, non-linear relationships, and noise) but that does not actually include any real patient data. While previous research has explored models for generating synthetic data sets, here we explore the integration of resampling, probabilistic graphical modelling, latent variable identification, and outlier analysis for producing realistic synthetic data based on UK primary care patient data. In particular, we focus on handling missingness, complex interactions between variables, and the resulting sensitivity analysis statistics from machine learning classifiers, while quantifying the risks of patient re-identification from synthetic datapoints. We show that, through our approach of integrating outlier analysis with graphical modelling and resampling, we can achieve synthetic data sets that are not significantly different from original ground truth data in terms of feature distributions, feature dependencies, and sensitivity analysis statistics when inferring machine learning classifiers. What is more, the risk of generating synthetic data that is identical or very similar to real patients is shown to be low.

Using topological data analysis and pseudo time series to infer temporal phenotypes from electronic health records.

Temporal phenotyping enables clinicians to better understand observable characteristics of a disease as it progresses. Modelling disease progression that captures interactions between phenotypes is inherently challenging. Temporal models that capture change in disease over time can identify the key features that characterize disease subtypes that underpin these trajectories. These models will enable clinicians to identify early warning signs of progression in specific sub-types and therefore to make informed decisions tailored to individual patients. In this paper, we explore two approaches to building temporal phenotypes based on the topology of data: topological data analysis and pseudo time-series. Using type 2 diabetes data, we show that the topological data analysis approach is able to identify disease trajectories and that pseudo time-series can infer a state space model characterized by transitions between hidden states that represent distinct temporal phenotypes. Both approaches highlight lipid profiles as key factors in distinguishing the phenotypes.

Consensus and Meta-analysis regulatory networks for combining multiple microarray gene expression datasets.

Microarray data is a key source of experimental data for modelling gene regulatory interactions from expression levels. With the rapid increase of publicly available microarray data comes the opportunity to produce regulatory network models based on multiple datasets. Such models are potentially more robust with greater confidence, and place less reliance on a single dataset. However, combining datasets directly can be difficult as experiments are often conducted on different microarray platforms, and in different laboratories leading to inherent biases in the data that are not always removed through pre-processing such as normalisation. In this paper we compare two frameworks for combining microarray datasets to model regulatory networks: pre- and post-learning aggregation. In pre-learning approaches, such as using simple scale-normalisation prior to the concatenation of datasets, a model is learnt from a combined dataset, whilst in post-learning aggregation individual models are learnt from each dataset and the models are combined. We present two novel approaches for post-learning aggregation, each based on aggregating high-level features of Bayesian network models that have been generated from different microarray expression datasets. Meta-analysis Bayesian networks are based on combining statistical confidences attached to network edges whilst Consensus Bayesian networks identify consistent network features across all datasets. We apply both approaches to multiple datasets from synthetic and real (Escherichia coli and yeast) networks and demonstrate that both methods can improve on networks learnt from a single dataset or an aggregated dataset formed using a standard scale-normalisation.

Nearest Consensus Clustering Classification to Identify Subclasses and Predict Disease.

Disease subtyping, which helps to develop personalized treatments, remains a challenge in data analysis because of the many different ways to group patients based upon their data. However, if we can identify subclasses of disease, then it will help to develop better models that are more specific to individuals and should therefore improve prediction and understanding of the underlying characteristics of the disease in question. This paper proposes a new algorithm that integrates consensus clustering methods with classification in order to overcome issues with sample bias. The new algorithm combines K-means with consensus clustering in order build cohort-specific decision trees that improve classification as well as aid the understanding of the underlying differences of the discovered groups. The methods are tested on a real-world freely available breast cancer dataset and data from a London hospital on systemic sclerosis, a rare potentially fatal condition. Results show that "nearest consensus clustering classification" improves the accuracy and the prediction significantly when this algorithm has been compared with competitive similar methods.

Construction of a network describing asparagine metabolism in plants and its application to the identification of genes affecting asparagine metabolism in wheat under drought and nutritional stress.

A detailed network describing asparagine metabolism in plants was constructed using published data from Arabidopsis (Arabidopsis thaliana) maize (Zea mays), wheat (Triticum aestivum), pea (Pisum sativum), soybean (Glycine max), lupin (Lupus albus), and other species, including animals. Asparagine synthesis and degradation is a major part of amino acid and nitrogen metabolism in plants. The complexity of its metabolism, including limiting and regulatory factors, was represented in a logical sequence in a pathway diagram built using yED graph editor software. The network was used with a Unique Network Identification Pipeline in the analysis of data from 18 publicly available transcriptomic data studies. This identified links between genes involved in asparagine metabolism in wheat roots under drought stress, wheat leaves under drought stress, and wheat leaves under conditions of sulfur and nitrogen deficiency. The network represents a powerful aid for interpreting the interactions not only between the genes in the pathway but also among enzymes, metabolites and smaller molecules. It provides a concise, clear understanding of the complexity of asparagine metabolism that could aid the interpretation of data relating to wider amino acid metabolism and other metabolic processes.

Experts integrate explicit contextual priors and environmental information to improve anticipation efficiency.

The understanding of how experts integrate prior situation-specific information (i.e., contextual priors) with emergent visual information when performing dynamic and temporally constrained tasks is limited. We used a soccer-based anticipation task to examine the ability of expert and novice players to integrate prior information about an opponent's action tendencies with unfolding environmental information such as opponent kinematics. We recorded gaze behaviors and ongoing expectations during task performance. Moreover, we assessed their final anticipatory judgments and perceived levels of cognitive effort invested. Explicit contextual priors biased the allocation of visual attention and shaped ongoing expectations in experts but not in novices. When the final action was congruent with the most likely action given the opponent's action tendencies, the contextual priors enhanced the final judgments for both groups. For incongruent trials, the explicit priors had a negative impact on the final judgments of novices but not experts. We interpreted the data using a Bayesian framework to provide novel insights into how contextual priors and dynamic environmental information are combined when making decisions under time pressure. Moreover, we provide evidence that this integration is governed by the temporal relevance of the information at hand as well as the ability to infer this relevance. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

Reprint of "Updating Markov models to integrate cross-sectional and longitudinal studies".

Clinical trials are typically conducted over a population within a defined time period in order to illuminate certain characteristics of a health issue or disease process. Cross-sectional studies provide a snapshot of these disease processes over a large number of people but do not allow us to model the temporal nature of disease, which is essential for modelling detailed prognostic predictions. Longitudinal studies, on the other hand, are used to explore how these processes develop over time in a number of people but can be expensive and time-consuming, and many studies only cover a relatively small window within the disease process. This paper explores the application of intelligent data analysis techniques for building reliable models of disease progression from both cross-sectional and longitudinal studies. The aim is to learn disease 'trajectories' from cross-sectional data by building realistic trajectories from healthy patients to those with advanced disease. We focus on exploring whether we can 'calibrate' models learnt from these trajectories with real longitudinal data using Baum-Welch re-estimation so that the dynamic parameters reflect the true underlying processes more closely. We use Kullback-Leibler distance and Wilcoxon rank metrics to assess how calibration improves the models to better reflect the underlying dynamics.

Updating Markov models to integrate cross-sectional and longitudinal studies.

Clinical trials are typically conducted over a population within a defined time period in order to illuminate certain characteristics of a health issue or disease process. Cross-sectional studies provide a snapshot of these disease processes over a large number of people but do not allow us to model the temporal nature of disease, which is essential for modelling detailed prognostic predictions. Longitudinal studies, on the other hand, are used to explore how these processes develop over time in a number of people but can be expensive and time-consuming, and many studies only cover a relatively small window within the disease process. This paper explores the application of intelligent data analysis techniques for building reliable models of disease progression from both cross-sectional and longitudinal studies. The aim is to learn disease 'trajectories' from cross-sectional data by building realistic trajectories from healthy patients to those with advanced disease. We focus on exploring whether we can 'calibrate' models learnt from these trajectories with real longitudinal data using Baum-Welch re-estimation so that the dynamic parameters reflect the true underlying processes more closely. We use Kullback-Leibler distance and Wilcoxon rank metrics to assess how calibration improves the models to better reflect the underlying dynamics.

Structure and function in glaucoma: The relationship between a functional visual field map and an anatomic retinal map.

PURPOSE: To examine the relationship between an anatomic map relating the retinal nerve fiber layer (RNFL) distribution to the optic nerve head and a functional map derived from the interpoint correlation of raw sensitivities in visual field (VF) testing. METHODS: Previously, interpoint correlations were generated for all possible pairs of VF test points in a dataset of 98,821 Humphrey VF test results taken from the Moorfields Eye Hospital archive. The relationship between these correlations and the physical distance between the VF test point pairs was evaluated by Pearson's correlation coefficient and multiple regression analysis. The distance between the pairs of VF test points was calculated in two ways. First, the anatomic map was used to estimate the angular distance at the optic nerve head (ONH), between the RNFL bundles corresponding to the VF test points in each pair (ONHd). Second, the retinal distance between pairs of test points was calculated from the Humphrey VF template (RETd). A best-fit model for predicting functional correlation (FC) from ONHd and RETd was constructed and used to formulate a filter incorporating the anatomic-functional correlation data. RESULTS: All scatterplots showed a negative association between interpoint retinal sensitivity correlation values and distance between points: ONHd (R2 = 0.60) and RETd (R2 = 0.33). The raw sensitivity correlation values could be predicted from a multiple regression model using ONHd, RETd, and a combined interaction of ONHd and RETd (R2 = 0.75, P < 0.00001). The construction of a new filter was based on the equation FC = 0.9325 - (0.0029 . ONHd) - (0.0077 . RETd) + (0.0001 . ONHd . RETd). CONCLUSIONS: A good level of association was observed between the strength of correlation between points in the VF and the relative location of those test points in the peripheral retina and in corresponding RNFL bundles at the ONH. These results help to validate the relationship between structure and function and may be of use in the further refinement of physiologically derived VF filters to reduce measurement noise.