Organoids have immense potential as ex vivo disease models for drug discovery and personalized drug screening. Dynamic changes in individual organoid morphology, number, and size can indicate important drug responses. However, these metrics are difficult and labor-intensive to obtain for high-throughput image datasets. Here, we present OrganoID, a robust image analysis platform that automatically recognizes, labels, and tracks single organoids, pixel-by-pixel, in brightfield and phase-contrast microscopy experiments. The platform was trained on images of pancreatic cancer organoids and validated on separate images of pancreatic, lung, colon, and adenoid cystic carcinoma organoids, which showed excellent agreement with manual measurements of organoid count (95%) and size (97%) without any parameter adjustments. Single-organoid tracking accuracy remained above 89% over a four-day time-lapse microscopy study. Automated single-organoid morphology analysis of a chemotherapy dose-response experiment identified strong dose effect sizes on organoid circularity, solidity, and eccentricity. OrganoID enables straightforward, detailed, and accurate image analysis to accelerate the use of organoids in high-throughput, data-intensive biomedical applications.

With the rapid development of AI technology in recent years, there have been many studies with deep learning models in soft sensing area. However, the models have become more and more complex, yet, the data sets remain limited: researchers are fitting million-parameter models with hundreds of data samples, which is insufficient to exercise the effectiveness of their models. To solve this long-lasting problem, we are providing large-scale high-dimensional time series manufacturing sensor data from Seagate Technology to the public. We demonstrate the challenges and effectiveness of modeling industrial big data by a Soft Sensing Transformer model on these data sets. Transformer is used because, it has outperformed state-of-the-art techniques in Natural Language Processing, and since then has also performed well in the direct application to computer vision without introduction of image- specific inductive biases. We observe the similarity of a sentence structure to the sensor readings and process the multi-variable sensor readings in a similar manner of sentences in natural language. The high-dimensional time series data is formatted into the same shape of embedded sentences and fed into the transformer model. The results show that transformer model outperforms the benchmark models in soft sensing field based on auto-encoder and long short-term memory (LSTM). To the best of our knowledge, we are the first team in academia or industry to benchmark the performance of original transformer model with large-scale numerical soft sensing data. Additionally, In contrast to the natural language processing or computer vision tasks where human-level performances are regarded as golden standards, our large-scale soft sensing study is an example that transformer is able to interpret high-dimensional numerical data which is not interpretable by human.

The human sex ratio at birth (SRB), defined as the ratio between the number of newborn boys to the total number of newborns, is typically slightly greater than 1/2 (more boys than girls) and tends to vary across different geographical regions and time periods. In this large-scale study, we sought to validate previously-reported associations and test new hypotheses using statistical analysis of two very large datasets incorporating electronic medical records (EMRs). One of the datasets represents over half (∼ 150 million) of the US population for over 8 years (IBM Watson Health MarketScan insurance claims) while another covers the entire Swedish population (∼ 9 million) for over 30 years (the Swedish National Patient Register). After testing more than 100 hypotheses, we showed that neither dataset supported models in which the SRB changed seasonally or in response to variations in ambient temperature. However, increased levels of a diverse array of air and water pollutants, were associated with lower SRBs, including increased levels of industrial and agricultural activity, which served as proxies for water pollution. Moreover, some exogenous factors generally considered to be environmental toxins turned out to induce higher SRBs. Finally, we identified new factors with signals for either higher or lower SRBs. In all cases, the effect sizes were modest but highly statistically significant owing to the large sizes of the two datasets. We suggest that while it was unlikely that the associations have arisen from sex-specific selection mechanisms, they are still useful for the purpose of public health surveillance if they can be corroborated by empirical evidences.

The transition to daylight saving time (DST) is beneficial for energy conservation but at the same time it has been reported to increase the risk of cerebrovascular and cardiovascular problems. Here, we evaluate the effect of the DST shift on a whole spectrum of diseases-an analysis we hope will be helpful in weighing the risks and benefits of DST shifts. Our study relied on a population-based, cross-sectional analysis of the IBM Watson Health MarketScan insurance claim dataset, which incorporates over 150 million unique patients in the US, and the Swedish national inpatient register, which incorporates more than nine million unique Swedes. For hundreds of sex- and age-specific diseases, we assessed effects of the DST shifts forward and backward by one hour in spring and autumn by comparing the observed and expected diagnosis rates after DST shift exposure. We found four prominent, elevated risk clusters, including cardiovascular diseases (such as heart attacks), injuries, mental and behavioral disorders, and immune-related diseases such as noninfective enteritis and colitis to be significantly associated with DST shifts in the United States and Sweden. While the majority of disease risk elevations are modest (a few percent), a considerable number of diseases exhibit an approximately ten percent relative risk increase. We estimate that each spring DST shift is associated with negative health effects-with 150,000 incidences in the US, and 880,000 globally. We also identify for the first time a collection of diseases with relative risks that appear to decrease immediately after the spring DST shift, enriched with infections and immune system-related maladies. These diseases' decreasing relative risks might be driven by the documented boosting effect of a short-term stress (such as that experienced around the spring DST shift) on the immune system.

Anonymized electronic medical records are an increasingly popular source of research data. However, these datasets often lack race and ethnicity information. This creates problems for researchers modeling human disease, as race and ethnicity are powerful confounders for many health exposures and treatment outcomes; race and ethnicity are closely linked to population-specific genetic variation. We showed that deep neural networks generate more accurate estimates for missing racial and ethnic information than competing methods (e.g., logistic regression, random forest, support vector machines, and gradient-boosted decision trees). RIDDLE yielded significantly better classification performance across all metrics that were considered: accuracy, cross-entropy loss (error), precision, recall, and area under the curve for receiver operating characteristic plots (all p < 10-9). We made specific efforts to interpret the trained neural network models to identify, quantify, and visualize medical features which are predictive of race and ethnicity. We used these characterizations of informative features to perform a systematic comparison of differential disease patterns by race and ethnicity. The fact that clinical histories are informative for imputing race and ethnicity could reflect (1) a skewed distribution of blue- and white-collar professions across racial and ethnic groups, (2) uneven accessibility and subjective importance of prophylactic health, (3) possible variation in lifestyle, such as dietary habits, and (4) differences in background genetic variation which predispose to diseases.

Synonymous relationships among biomedical terms are extensively annotated within specialized terminologies, implying that synonymy is important for practical computational applications within this field. It remains unclear, however, whether text mining actually benefits from documented synonymy and whether existing biomedical thesauri provide adequate coverage of these linguistic relationships. In this study, we examine the impact and extent of undocumented synonymy within a very large compendium of biomedical thesauri. First, we demonstrate that missing synonymy has a significant negative impact on named entity normalization, an important problem within the field of biomedical text mining. To estimate the amount synonymy currently missing from thesauri, we develop a probabilistic model for the construction of synonym terminologies that is capable of handling a wide range of potential biases, and we evaluate its performance using the broader domain of near-synonymy among general English words. Our model predicts that over 90% of these relationships are currently undocumented, a result that we support experimentally through "crowd-sourcing." Finally, we apply our model to biomedical terminologies and predict that they are missing the vast majority (>90%) of the synonymous relationships they intend to document. Overall, our results expose the dramatic incompleteness of current biomedical thesauri and suggest the need for "next-generation," high-coverage lexical terminologies.

Many factors affect the risks for neurodevelopmental maladies such as autism spectrum disorders (ASD) and intellectual disability (ID). To compare environmental, phenotypic, socioeconomic and state-policy factors in a unified geospatial framework, we analyzed the spatial incidence patterns of ASD and ID using an insurance claims dataset covering nearly one third of the US population. Following epidemiologic evidence, we used the rate of congenital malformations of the reproductive system as a surrogate for environmental exposure of parents to unmeasured developmental risk factors, including toxins. Adjusted for gender, ethnic, socioeconomic, and geopolitical factors, the ASD incidence rates were strongly linked to population-normalized rates of congenital malformations of the reproductive system in males (an increase in ASD incidence by 283% for every percent increase in incidence of malformations, 95% CI: [91%, 576%], p<6×10(-5)). Such congenital malformations were barely significant for ID (94% increase, 95% CI: [1%, 250%], p = 0.0384). Other congenital malformations in males (excluding those affecting the reproductive system) appeared to significantly affect both phenotypes: 31.8% ASD rate increase (CI: [12%, 52%], p<6×10(-5)), and 43% ID rate increase (CI: [23%, 67%], p<6×10(-5)). Furthermore, the state-mandated rigor of diagnosis of ASD by a pediatrician or clinician for consideration in the special education system was predictive of a considerable decrease in ASD and ID incidence rates (98.6%, CI: [28%, 99.99%], p = 0.02475 and 99% CI: [68%, 99.99%], p = 0.00637 respectively). Thus, the observed spatial variability of both ID and ASD rates is associated with environmental and state-level regulatory factors; the magnitude of influence of compound environmental predictors was approximately three times greater than that of state-level incentives. The estimated county-level random effects exhibited marked spatial clustering, strongly indicating existence of as yet unidentified localized factors driving apparent disease incidence. Finally, we found that the rates of ASD and ID at the county level were weakly but significantly correlated (Pearson product-moment correlation 0.0589, p = 0.00101), while for females the correlation was much stronger (0.197, p<2.26×10(-16)).

Text mining applications typically involve statistical models that require accessing and updating model parameters in an iterative fashion. With the growing size of the data, such models become extremely parameter rich, and naive parallel implementations fail to address the scalability problem of maintaining a distributed look-up table that maps model parameters to their values. We evaluate several existing alternatives to provide coordination among worker nodes in Hadoop [11] clusters, and suggest a new multi-layered look-up architecture that is specifically optimized for certain problem domains. Our solution exploits the power-law distribution characteristics of the phrase or n-gram counts in large corpora while utilizing a Bloom Filter [2], in-memory cache, and an HBase [12] cluster at varying levels of abstraction.

We study text analysis algorithms that use global optimization methods to compute local characteristics that are consistent with properties of the entire corpus rather than computed locally based on exogenous parameters. In the iterative implementations that we consider, each step both reads and updates a database of parameter values. Motivated by a need for rapid analysis of large corpora, we have developed methods for efficient access to such databases on parallel computers. These methods combine Bloom filters, in-memory caches, and an HBase cluster to reduce communication costs greatly relative to simpler approaches that either fully distribute or fully replicate the database. We also describe how this method can be incorporated into the MapReduce programming model, and illustrate its use within phrase segmentation programs. Our design can achieve considerable run time, latency and storage space improvements relative to other methods. In one phrase segmentation application, we improve performance by a factor of six relative to an HBase-based implementation.

Computational models in biomedicine rely on biological and clinical assumptions. The selection of these assumptions contributes substantially to modeling success or failure. Assumptions used by experts at the cutting edge of research, however, are rarely explicitly described in scientific publications. One can directly collect and assess some of these assumptions through interviews and surveys. Here we investigate diversity in expert views about a complex biological phenomenon, the process of cancer metastasis. We harvested individual viewpoints from 28 experts in clinical and molecular aspects of cancer metastasis and summarized them computationally. While experts predominantly agreed on the definition of individual steps involved in metastasis, no two expert scenarios for metastasis were identical. We computed the probability that any two experts would disagree on k or fewer metastatic stages and found that any two randomly selected experts are likely to disagree about several assumptions. Considering the probability that two or more of these experts review an article or a proposal about metastatic cascades, the probability that they will disagree with elements of a proposed model approaches 1. This diversity of conceptions has clear consequences for advance and deadlock in the field. We suggest that strong, incompatible views are common in biomedicine but largely invisible to biomedical experts themselves. We built a formal Markov model of metastasis to encapsulate expert convergence and divergence regarding the entire sequence of metastatic stages. This model revealed stages of greatest disagreement, including the points at which cancer enters and leaves the bloodstream. The model provides a formal probabilistic hypothesis against which researchers can evaluate data on the process of metastasis. This would enable subsequent improvement of the model through Bayesian probabilistic update. Practically, we propose that model assumptions and hunches be harvested systematically and made available for modelers and scientists.

The use of structured knowledge representations-ontologies and terminologies-has become standard in biomedicine. Definitions of ontologies vary widely, as do the values and philosophies that underlie them. In seeking to make these views explicit, we conducted and summarized interviews with a dozen leading ontologists. Their views clustered into three broad perspectives that we summarize as mathematics, computer code, and Esperanto. Ontology as mathematics puts the ultimate premium on rigor and logic, symmetry and consistency of representation across scientific subfields, and the inclusion of only established, non-contradictory knowledge. Ontology as computer code focuses on utility and cultivates diversity, fitting ontologies to their purpose. Like computer languages C++, Prolog, and HTML, the code perspective holds that diverse applications warrant custom designed ontologies. Ontology as Esperanto focuses on facilitating cross-disciplinary communication, knowledge cross-referencing, and computation across datasets from diverse communities. We show how these views align with classical divides in science and suggest how a synthesis of their concerns could strengthen the next generation of biomedical ontologies.

A scientific ontology is a formal representation of knowledge within a domain, typically including central concepts, their properties, and relations. With the rise of computers and high-throughput data collection, ontologies have become essential to data mining and sharing across communities in the biomedical sciences. Powerful approaches exist for testing the internal consistency of an ontology, but not for assessing the fidelity of its domain representation. We introduce a family of metrics that describe the breadth and depth with which an ontology represents its knowledge domain. We then test these metrics using (1) four of the most common medical ontologies with respect to a corpus of medical documents and (2) seven of the most popular English thesauri with respect to three corpora that sample language from medicine, news, and novels. Here we show that our approach captures the quality of ontological representation and guides efforts to narrow the breach between ontology and collective discourse within a domain. Our results also demonstrate key features of medical ontologies, English thesauri, and discourse from different domains. Medical ontologies have a small intersection, as do English thesauri. Moreover, dialects characteristic of distinct domains vary strikingly as many of the same words are used quite differently in medicine, news, and novels. As ontologies are intended to mirror the state of knowledge, our methods to tighten the fit between ontology and domain will increase their relevance for new areas of biomedical science and improve the accuracy and power of inferences computed across them.

In the biomedical domain verbs tend to be drawn from a smaller and more regular vocabulary than standard English and are especially informative for predicting the semantic class of the entities involved. With this motivation we investigate the discovery and counting of verbs in the biomedical domain by applying part-of-speech tagging and unsupervised morphological analysis. Our goal is to automatically discover an almost complete set of relevant verbs and group lexical variants of the same verb into semantic classes. Additionally, we analyze differences in verb usage between biomedical and standard English, and between general biomedical texts and a specific sub domain. This verb frequency data can serve as a basis for named entity recognition in biomedical texts.

We have generated and made publicly available two very large networks of molecular interactions: 49,493 mouse-specific and 52,518 human-specific interactions. These networks were generated through automated analysis of 368,331 full-text research articles and 8,039,972 article abstracts from the PubMed database, using the GeneWays system. Our networks cover a wide spectrum of molecular interactions, such as bind, phosphorylate, glycosylate, and activate; 207 of these interaction types occur more than 1,000 times in our unfiltered, multi-species data set. Because mouse and human genes are linked through an orthological relationship, human and mouse networks are amenable to straightforward, joint computational analysis. Using our newly generated networks and known associations between mouse genes and cerebellar malformation phenotypes, we predicted a number of new associations between genes and five cerebellar phenotypes (small cerebellum, absent cerebellum, cerebellar degeneration, abnormal foliation, and abnormal vermis). Using a battery of statistical tests, we showed that genes that are associated with cerebellar phenotypes tend to form compact network clusters. Further, we observed that cerebellar malformation phenotypes tend to be associated with highly connected genes. This tendency was stronger for developmental phenotypes and weaker for cerebellar degeneration.

No abstract

Large-scale annotation efforts typically involve several experts who may disagree with each other. We propose an approach for modeling disagreements among experts that allows providing each annotation with a confidence value (i.e., the posterior probability that it is correct). Our approach allows computing certainty-level for individual annotations, given annotator-specific parameters estimated from data. We developed two probabilistic models for performing this analysis, compared these models using computer simulation, and tested each model's actual performance, based on a large data set generated by human annotators specifically for this study. We show that even in the worst-case scenario, when all annotators disagree, our approach allows us to significantly increase the probability of choosing the correct annotation. Along with this publication we make publicly available a corpus of 10,000 sentences annotated according to several cardinal dimensions that we have introduced in earlier work. The 10,000 sentences were all 3-fold annotated by a group of eight experts, while a 1,000-sentence subset was further 5-fold annotated by five new experts. While the presented data represent a specialized curation task, our modeling approach is general; most data annotation studies could benefit from our methodology.

MOTIVATION: Much current research in biomedical text mining is concerned with serving biologists by extracting certain information from scientific text. We note that there is no 'average biologist' client; different users have distinct needs. For instance, as noted in past evaluation efforts (BioCreative, TREC, KDD) database curators are often interested in sentences showing experimental evidence and methods. Conversely, lab scientists searching for known information about a protein may seek facts, typically stated with high confidence. Text-mining systems can target specific end-users and become more effective, if the system can first identify text regions rich in the type of scientific content that is of interest to the user, retrieve documents that have many such regions, and focus on fact extraction from these regions. Here, we study the ability to characterize and classify such text automatically. We have recently introduced a multi-dimensional categorization and annotation scheme, developed to be applicable to a wide variety of biomedical documents and scientific statements, while intended to support specific biomedical retrieval and extraction tasks. RESULTS: The annotation scheme was applied to a large corpus in a controlled effort by eight independent annotators, where three individual annotators independently tagged each sentence. We then trained and tested machine learning classifiers to automatically categorize sentence fragments based on the annotation. We discuss here the issues involved in this task, and present an overview of the results. The latter strongly suggest that automatic annotation along most of the dimensions is highly feasible, and that this new framework for scientific sentence categorization is applicable in practice.

BACKGROUND: While biomedical text mining is emerging as an important research area, practical results have proven difficult to achieve. We believe that an important first step towards more accurate text-mining lies in the ability to identify and characterize text that satisfies various types of information needs. We report here the results of our inquiry into properties of scientific text that have sufficient generality to transcend the confines of a narrow subject area, while supporting practical mining of text for factual information. Our ultimate goal is to annotate a significant corpus of biomedical text and train machine learning methods to automatically categorize such text along certain dimensions that we have defined. RESULTS: We have identified five qualitative dimensions that we believe characterize a broad range of scientific sentences, and are therefore useful for supporting a general approach to text-mining: focus, polarity, certainty, evidence, and directionality. We define these dimensions and describe the guidelines we have developed for annotating text with regard to them. To examine the effectiveness of the guidelines, twelve annotators independently annotated the same set of 101 sentences that were randomly selected from current biomedical periodicals. Analysis of these annotations shows 70-80% inter-annotator agreement, suggesting that our guidelines indeed present a well-defined, executable and reproducible task. CONCLUSION: We present our guidelines defining a text annotation task, along with annotation results from multiple independently produced annotations, demonstrating the feasibility of the task. The annotation of a very large corpus of documents along these guidelines is currently ongoing. These annotations form the basis for the categorization of text along multiple dimensions, to support viable text mining for experimental results, methodology statements, and other forms of information. We are currently developing machine learning methods, to be trained and tested on the annotated corpus, that would allow for the automatic categorization of biomedical text along the general dimensions that we have presented. The guidelines in full detail, along with annotated examples, are publicly available.

Text-mining algorithms make mistakes in extracting facts from natural-language texts. In biomedical applications, which rely on use of text-mined data, it is critical to assess the quality (the probability that the message is correctly extracted) of individual facts--to resolve data conflicts and inconsistencies. Using a large set of almost 100,000 manually produced evaluations (most facts were independently reviewed more than once, producing independent evaluations), we implemented and tested a collection of algorithms that mimic human evaluation of facts provided by an automated information-extraction system. The performance of our best automated classifiers closely approached that of our human evaluators (ROC score close to 0.95). Our hypothesis is that, were we to use a larger number of human experts to evaluate any given sentence, we could implement an artificial-intelligence curator that would perform the classification job at least as accurately as an average individual human evaluator. We illustrated our analysis by visualizing the predicted accuracy of the text-mined relations involving the term cocaine.

Mark Borodovsky, Andrey Rzhetsky, Dmitrij Frishman; The Fifth Georgia Tech—Oak Ridge National Laboratory International Conference in Bioinformatics: in sil

Information on molecular networks, such as networks of interacting proteins, comes from diverse sources that contain remarkable differences in distribution and quantity of errors. Here, we introduce a probabilistic model useful for predicting protein interactions from heterogeneous data sources. The model describes stochastic generation of protein-protein interaction networks with real-world properties, as well as generation of two heterogeneous sources of protein-interaction information: research results automatically extracted from the literature and yeast two-hybrid experiments. Based on the domain composition of proteins, we use the model to predict protein interactions for pairs of proteins for which no experimental data are available. We further explore the prediction limits, given experimental data that cover only part of the underlying protein networks. This approach can be extended naturally to include other types of biological data sources.

In order to understand the molecular machinery of the cell, we need to know about the multitude of protein-protein interactions that allow the cell to function. High-throughput technologies provide some data about these interactions, but so far that data is fairly noisy. Therefore, computational techniques for predicting protein-protein interactions could be of significant value. One approach to predicting interactions in silico is to produce from first principles a detailed model of a candidate interaction. We take an alternative approach, employing a relatively simple model that learns dynamically from a large collection of data. In this work, we describe an attraction-repulsion model, in which the interaction between a pair of proteins is represented as the sum of attractive and repulsive forces associated with small, domain- or motif-sized features along the length of each protein. The model is discriminative, learning simultaneously from known interactions and from pairs of proteins that are known (or suspected) not to interact. The model is efficient to compute and scales well to very large collections of data. In a cross-validated comparison using known yeast interactions, the attraction-repulsion method performs better than several competing techniques.

Knowledge on interactions between molecules in living cells is indispensable for theoretical analysis and practical applications in modern genomics and molecular biology. Building such networks relies on the assumption that the correct molecular interactions are known or can be identified by reading a few research articles. However, this assumption does not necessarily hold, as truth is rather an emerging property based on many potentially conflicting facts. This paper explores the processes of knowledge generation and publishing in the molecular biology literature using modelling and analysis of real molecular interaction data. The data analysed in this article were automatically extracted from 50000 research articles in molecular biology using a computer system called GeneWays containing a natural language processing module. The paper indicates that truthfulness of statements is associated in the minds of scientists with the relative importance (connectedness) of substances under study, revealing a potential selection bias in the reporting of research results. Aiming at understanding the statistical properties of the life cycle of biological facts reported in research articles, we formulate a stochastic model describing generation and propagation of knowledge about molecular interactions through scientific publications. We hope that in the future such a model can be useful for automatically producing consensus views of molecular interaction data.

MOTIVATION: Current growth in the field of genomics has provided a number of exciting approaches to the modeling of evolutionary mechanisms within the genome. Separately, dynamical and statistical analyses of networks such as the World Wide Web and the social interactions existing between humans have shown that these networks can exhibit common fractal properties-including the property of being scale-free. This work attempts to bridge these two fields and demonstrate that the fractal properties of molecular networks are linked to the fractal properties of their underlying genomes. RESULTS: We suggest a stochastic model capable of describing the evolutionary growth of metabolic or signal-transduction networks. This model generates networks that share important statistical properties (so-called scale-free behavior) with real molecular networks. In particular, the frequency of vertices connected to exactly k other vertices follows a power-law distribution. The shape of this distribution remains invariant to changes in network scale: a small subgraph has the same distribution as the complete graph from which it is derived. Furthermore, the model correctly predicts that the frequencies of distinct DNA and protein domains also follow a power-law distribution. Finally, the model leads to a simple equation linking the total number of different DNA and protein domains in a genome with both the total number of genes and the overall network topology. AVAILABILITY: MatLab (MathWorks, Inc.) programs described in this manuscript are available on request from the authors. CONTACT: [email protected].

MOTIVATION: In order to aid in hypothesis-driven experimental gene discovery, we are designing a computer application for the automatic retrieval of signal transduction data from electronic versions of scientific publications using natural language processing (NLP) techniques, as well as for visualizing and editing representations of regulatory systems. These systems describe both signal transduction and biochemical pathways within complex multicellular organisms, yeast, and bacteria. This computer application in turn requires the development of a domain-specific ontology, or knowledge model. RESULTS: We introduce an ontological model for the representation of biological knowledge related to regulatory networks in vertebrates. We outline a taxonomy of the concepts, define their 'whole-to-part' relationships, describe the properties of major concepts, and outline a set of the most important axioms. The ontology is partially realized in a computer system designed to aid researchers in biology and medicine in visualizing and editing a representation of a signal transduction system.

The METREE program package for estimating phylogenetic trees with the minimum evolution method is written in Turbo C 2.0 and is intended to be used on any IBM-compatible personal computers that have a mathematical coprocessor. The package is simple to use and is menu driven. A program for visualizing and printing out the final tree is also included.