Continuous-Time and Dynamic Suicide Attempt Risk Prediction with Neural Ordinary Differential Equations.

Suicide is one of the leading causes of death in the US, and the number of attributable deaths continues to increase. Risk of suicide-related behaviors (SRBs) is dynamic, and SRBs can occur across a continuum of time and locations. However, current SRB risk assessment methods, whether conducted by clinicians or through machine learning models, treat SRB risk as static and are confined to specific times and locations, such as following a hospital visit. Such a paradigm is unrealistic as SRB risk fluctuates and creates time gaps in the availability of risk scores. Here, we develop two closely related model classes, Event-GRU-ODE and Event-GRU-Discretized, that can predict the dynamic risk of events as a continuous trajectory based on Neural ODEs, an advanced AI model class for time series prediction. As such, these models can estimate changes in risk across the continuum of future time points, even without new observations, and can update these estimations as new data becomes available. We train and validate these models for SRB prediction using a large electronic health records database. Both models demonstrated high discrimination performance for SRB prediction (e.g., AUROC > 0.92 in the full, general cohort), serving as an initial step toward developing novel and comprehensive suicide prevention strategies based on dynamic changes in risk.

MELLODDY: Cross-pharma Federated Learning at Unprecedented Scale Unlocks Benefits in QSAR without Compromising Proprietary Information.

Federated multipartner machine learning has been touted as an appealing and efficient method to increase the effective training data volume and thereby the predictivity of models, particularly when the generation of training data is resource-intensive. In the landmark MELLODDY project, indeed, each of ten pharmaceutical companies realized aggregated improvements on its own classification or regression models through federated learning. To this end, they leveraged a novel implementation extending multitask learning across partners, on a platform audited for privacy and security. The experiments involved an unprecedented cross-pharma data set of 2.6+ billion confidential experimental activity data points, documenting 21+ million physical small molecules and 40+ thousand assays in on-target and secondary pharmacodynamics and pharmacokinetics. Appropriate complementary metrics were developed to evaluate the predictive performance in the federated setting. In addition to predictive performance increases in labeled space, the results point toward an extended applicability domain in federated learning. Increases in collective training data volume, including by means of auxiliary data resulting from single concentration high-throughput and imaging assays, continued to boost predictive performance, albeit with a saturating return. Markedly higher improvements were observed for the pharmacokinetics and safety panel assay-based task subsets.

Splitting chemical structure data sets for federated privacy-preserving machine learning.

With the increase in applications of machine learning methods in drug design and related fields, the challenge of designing sound test sets becomes more and more prominent. The goal of this challenge is to have a realistic split of chemical structures (compounds) between training, validation and test set such that the performance on the test set is meaningful to infer the performance in a prospective application. This challenge is by its own very interesting and relevant, but is even more complex in a federated machine learning approach where multiple partners jointly train a model under privacy-preserving conditions where chemical structures must not be shared between the different participating parties. In this work we discuss three methods which provide a splitting of a data set and are applicable in a federated privacy-preserving setting, namely: a. locality-sensitive hashing (LSH), b. sphere exclusion clustering, c. scaffold-based binning (scaffold network). For evaluation of these splitting methods we consider the following quality criteria (compared to random splitting): bias in prediction performance, classification label and data imbalance, similarity distance between the test and training set compounds. The main findings of the paper are a. both sphere exclusion clustering and scaffold-based binning result in high quality splitting of the data sets, b. in terms of compute costs sphere exclusion clustering is very expensive in the case of federated privacy-preserving setting.

A novel method for data fusion over entity-relation graphs and its application to protein-protein interaction prediction.

MOTIVATION: Modern bioinformatics is facing increasingly complex problems to solve, and we are indeed rapidly approaching an era in which the ability to seamlessly integrate heterogeneous sources of information will be crucial for the scientific progress. Here, we present a novel non-linear data fusion framework that generalizes the conventional matrix factorization paradigm allowing inference over arbitrary entity-relation graphs, and we applied it to the prediction of protein-protein interactions (PPIs). Improving our knowledge of PPI networks at the proteome scale is indeed crucial to understand protein function, physiological and disease states and cell life in general. RESULTS: We devised three data fusion-based models for the proteome-level prediction of PPIs, and we show that our method outperforms state of the art approaches on common benchmarks. Moreover, we investigate its predictions on newly published PPIs, showing that this new data has a clear shift in its underlying distributions and we thus train and test our models on this extended dataset. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

ChemGrapher: Optical Graph Recognition of Chemical Compounds by Deep Learning.

In drug discovery, knowledge of the graph structure of chemical compounds is essential. Many thousands of scientific articles and patents in chemistry and pharmaceutical sciences have investigated chemical compounds, but in many cases, the details of the structure of these chemical compounds are published only as an image. A tool to analyze these images automatically and convert them into a chemical graph structure would be useful for many applications, such as drug discovery. A few such tools are available and they are mostly derived from optical character recognition. However, our evaluation of the performance of these tools reveals that they often make mistakes in recognizing the correct bond multiplicity and stereochemical information. In addition, errors sometimes even lead to missing atoms in the resulting graph. In our work, we address these issues by developing a compound recognition method based on machine learning. More specifically, we develop a deep neural network model for optical compound recognition. The deep learning solution presented here consists of a segmentation model, followed by three classification models that predict atom locations, bonds, and charges. Furthermore, this model not only predicts the graph structure of the molecule but also provides all information necessary to relate each component of the resulting graph to the source image. This solution is scalable and can rapidly process thousands of images. Finally, we empirically compare the proposed method with the well-established tool OSRA1 and observe significant error reduction.

The complex microbiome from native semen to embryo culture environment in human in vitro fertilization procedure.

BACKGROUND: Only a few microbial studies have conducted in IVF (in vitro fertilization), showing the high-variety bacterial contamination of IVF culture media to cause damage to or even loss of cultured oocytes and embryos. We aimed to determine the prevalence and counts of bacteria in IVF samples, and to associate them with clinical outcome. METHODS: The studied samples from 50 infertile couples included: raw (n = 48), processed (n = 49) and incubated (n = 50) sperm samples, and IVF culture media (n = 50). The full microbiome was analyzed by 454 pyrosequencing and quantitative analysis by real-time quantitative PCR. Descriptive statistics, t-, Mann-Whitney tests and Spearman's correlation were used for comparison of studied groups. RESULTS: The study involved normozoospermic men. Normal vaginal microbiota was present in 72.0% of female partners, while intermediate microbiota and bacterial vaginosis were diagnosed in 12.0 and 16.0%, respectively. The decreasing bacterial loads were found in raw (35.5%), processed (12.0%) and sperm samples used for oocyte insemination (4.0%), and in 8.0% of IVF culture media. The most abundant genera of bacteria in native semen and IVF culture media were Lactobacillus, while in other samples Alphaproteobacteria prevailed. Staphylococcus sp. was found only in semen from patients with inflammation. Phylum Bacteroidetes was in negative correlation with sperm motility and Alphaproteobacteria with high-quality IVF embryos. CONCLUSION: Our study demonstrates that IVF does not occur in a sterile environment. The prevalent bacteria include classes Bacilli in raw semen and IVF culture media, Clostridia in processed and Bacteroidia in sperm samples used for insemination. The presence of Staphylococcus sp. and Alphaproteobacteria associated with clinical outcomes, like sperm and embryo quality.

An interpretable low-complexity machine learning framework for robust exome-based in-silico diagnosis of Crohn's disease patients.

Whole exome sequencing (WES) data are allowing researchers to pinpoint the causes of many Mendelian disorders. In time, sequencing data will be crucial to solve the genome interpretation puzzle, which aims at uncovering the genotype-to-phenotype relationship, but for the moment many conceptual and technical problems need to be addressed. In particular, very few attempts at the in-silico diagnosis of oligo-to-polygenic disorders have been made so far, due to the complexity of the challenge, the relative scarcity of the data and issues such as batch effects and data heterogeneity, which are confounder factors for machine learning (ML) methods. Here, we propose a method for the exome-based in-silico diagnosis of Crohn's disease (CD) patients which addresses many of the current methodological issues. First, we devise a rational ML-friendly feature representation for WES data based on the gene mutational burden concept, which is suitable for small sample sizes datasets. Second, we propose a Neural Network (NN) with parameter tying and heavy regularization, in order to limit its complexity and thus the risk of over-fitting. We trained and tested our NN on 3 CD case-controls datasets, comparing the performance with the participants of previous CAGI challenges. We show that, notwithstanding the limited NN complexity, it outperforms the previous approaches. Moreover, we interpret the NN predictions by analyzing the learned patterns at the variant and gene level and investigating the decision process leading to each prediction.

Author Correction: Linking drug target and pathway activation for effective therapy using multi-task learning.

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has been fixed in the paper.

GRNBoost2 and Arboreto: efficient and scalable inference of gene regulatory networks.

SUMMARY: Inferring a Gene Regulatory Network (GRN) from gene expression data is a computationally expensive task, exacerbated by increasing data sizes due to advances in high-throughput gene profiling technology, such as single-cell RNA-seq. To equip researchers with a toolset to infer GRNs from large expression datasets, we propose GRNBoost2 and the Arboreto framework. GRNBoost2 is an efficient algorithm for regulatory network inference using gradient boosting, based on the GENIE3 architecture. Arboreto is a computational framework that scales up GRN inference algorithms complying with this architecture. Arboreto includes both GRNBoost2 and an improved implementation of GENIE3, as a user-friendly open source Python package. AVAILABILITY AND IMPLEMENTATION: Arboreto is available under the 3-Clause BSD license at http://arboreto.readthedocs.io. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Towards practical privacy-preserving genome-wide association study.

BACKGROUND: The deployment of Genome-wide association studies (GWASs) requires genomic information of a large population to produce reliable results. This raises significant privacy concerns, making people hesitate to contribute their genetic information to such studies. RESULTS: We propose two provably secure solutions to address this challenge: (1) a somewhat homomorphic encryption (HE) approach, and (2) a secure multiparty computation (MPC) approach. Unlike previous work, our approach does not rely on adding noise to the input data, nor does it reveal any information about the patients. Our protocols aim to prevent data breaches by calculating the χ2 statistic in a privacy-preserving manner, without revealing any information other than whether the statistic is significant or not. Specifically, our protocols compute the χ2 statistic, but only return a yes/no answer, indicating significance. By not revealing the statistic value itself but only the significance, our approach thwarts attacks exploiting statistic values. We significantly increased the efficiency of our HE protocols by introducing a new masking technique to perform the secure comparison that is necessary for determining significance. CONCLUSIONS: We show that full-scale privacy-preserving GWAS is practical, as long as the statistics can be computed by low degree polynomials. Our implementations demonstrated that both approaches are efficient. The secure multiparty computation technique completes its execution in approximately 2 ms for data contributed by one million subjects.

Gene prioritization using Bayesian matrix factorization with genomic and phenotypic side information.

Motivation: Most gene prioritization methods model each disease or phenotype individually, but this fails to capture patterns common to several diseases or phenotypes. To overcome this limitation, we formulate the gene prioritization task as the factorization of a sparsely filled gene-phenotype matrix, where the objective is to predict the unknown matrix entries. To deliver more accurate gene-phenotype matrix completion, we extend classical Bayesian matrix factorization to work with multiple side information sources. The availability of side information allows us to make non-trivial predictions for genes for which no previous disease association is known. Results: Our gene prioritization method can innovatively not only integrate data sources describing genes, but also data sources describing Human Phenotype Ontology terms. Experimental results on our benchmarks show that our proposed model can effectively improve accuracy over the well-established gene prioritization method, Endeavour. In particular, our proposed method offers promising results on diseases of the nervous system; diseases of the eye and adnexa; endocrine, nutritional and metabolic diseases; and congenital malformations, deformations and chromosomal abnormalities, when compared to Endeavour. Availability and implementation: The Bayesian data fusion method is implemented as a Python/C++ package: https://github.com/jaak-s/macau. It is also available as a Julia package: https://github.com/jaak-s/BayesianDataFusion.jl. All data and benchmarks generated or analyzed during this study can be downloaded at https://owncloud.esat.kuleuven.be/index.php/s/UGb89WfkZwMYoTn. Supplementary information: Supplementary data are available at Bioinformatics online.

Linking drug target and pathway activation for effective therapy using multi-task learning.

Despite the abundance of large-scale molecular and drug-response data, the insights gained about the mechanisms underlying treatment efficacy in cancer has been in general limited. Machine learning algorithms applied to those datasets most often are used to provide predictions without interpretation, or reveal single drug-gene association and fail to derive robust insights. We propose to use Macau, a bayesian multitask multi-relational algorithm to generalize from individual drugs and genes and explore the interactions between the drug targets and signaling pathways' activation. A typical insight would be: "Activation of pathway Y will confer sensitivity to any drug targeting protein X". We applied our methodology to the Genomics of Drug Sensitivity in Cancer (GDSC) screening, using gene expression of 990 cancer cell lines, activity scores of 11 signaling pathways derived from the tool PROGENy as cell line input and 228 nominal targets for 265 drugs as drug input. These interactions can guide a tissue-specific combination treatment strategy, for example suggesting to modulate a certain pathway to maximize the drug response for a given tissue. We confirmed in literature drug combination strategies derived from our result for brain, skin and stomach tissues. Such an analysis of interactions across tissues might help target discovery, drug repurposing and patient stratification strategies.

Repurposing High-Throughput Image Assays Enables Biological Activity Prediction for Drug Discovery.

In both academia and the pharmaceutical industry, large-scale assays for drug discovery are expensive and often impractical, particularly for the increasingly important physiologically relevant model systems that require primary cells, organoids, whole organisms, or expensive or rare reagents. We hypothesized that data from a single high-throughput imaging assay can be repurposed to predict the biological activity of compounds in other assays, even those targeting alternate pathways or biological processes. Indeed, quantitative information extracted from a three-channel microscopy-based screen for glucocorticoid receptor translocation was able to predict assay-specific biological activity in two ongoing drug discovery projects. In these projects, repurposing increased hit rates by 50- to 250-fold over that of the initial project assays while increasing the chemical structure diversity of the hits. Our results suggest that data from high-content screens are a rich source of information that can be used to predict and replace customized biological assays.

The Influence of Different Maternal Microbial Communities on the Development of Infant Gut and Oral Microbiota.

Very few studies have analyzed how the composition of mother's microbiota affects the development of infant's gut and oral microbiota during the first months of life. Here, microbiota present in the mothers' gut, vagina, breast milk, oral cavity, and mammary areola were compared with the gut and oral microbiota of their infants over the first six months following birth. Samples were collected from the aforementioned body sites from seven mothers and nine infants at three different time points over a 6-month period. Each sample was analyzed with 16S rRNA gene sequencing. The gut microbiota of the infants harbored distinct microbial communities that had low similarity with the various maternal microbiota communities. In contrast, the oral microbiota of the infants exhibited high similarity with the microbiota of the mothers' breast milk, mammary areola and mouth. These results demonstrate that constant contact between microbial communities increases their similarity. A majority of the operational taxonomic units in infant gut and oral microbiota were also shared with the mothers' gut and oral communities, respectively. The disparity between the similarity and the proportion of the OTUs shared between infants' and mothers' gut microbiota might be related to lower diversity and therefore competition in infants' gut microbiota.

Near-Bottom Hypoxia Impacts Dynamics of Bacterioplankton Assemblage throughout Water Column of the Gulf of Finland (Baltic Sea).

Over the past century the spread of hypoxia in the Baltic Sea has been drastic, reaching its 'arm' into the easternmost sub-basin, the Gulf of Finland. The hydrographic and climatological properties of the gulf offer a broad suite of discrete niches for microbial communities. The current study explores spatiotemporal dynamics of bacterioplankton community in the Gulf of Finland using massively parallel sequencing of 16S rRNA fragments obtained by amplifying community DNA from spring to autumn period. The presence of redoxcline and drastic seasonal changes make spatiotemporal dynamics of bacterioplankton community composition (BCC) and abundances in such estuary remarkably complex. To the best of our knowledge, this is the first study that analyses spatiotemporal dynamics of BCC in relation to phytoplankton bloom throughout the water column (and redoxcline), not only at the surface layer. We conclude that capability to survive (or benefit from) shifts between oxic and hypoxic conditions is vital adaptation for bacteria to thrive in such environments. Our results contribute to the understanding of emerging patterns in BCCs that occupy hydrographically similar estuaries dispersed all over the world, and we suggest the presence of a global redox- and salinity-driven metacommunity. These results have important implications for understanding long-term ecological and biogeochemical impacts of hypoxia expansion in the Baltic Sea (and similar ecosystems), as well as global biogeography of bacteria specialized inhabiting similar ecosystems.

Endometrial transcriptome analysis indicates superiority of natural over artificial cycles in recurrent implantation failure patients undergoing frozen embryo transfer.

Little consensus has been reached on the best protocol for endometrial preparation for frozen embryo transfer (FET). It is not known how, and to what extent, hormone supplementation in artificial cycles influences endometrial preparation for embryo implantation at a molecular level, especially in patients who have experienced recurrent implantation failure. Transcriptome analysis of 15 endometrial biopsy samples at the time of embryo implantation was used to compare two different endometrial preparation protocols, natural versus artificial cycles, for FET in women who have experienced recurrent implantation failure compared with fertile women. IPA and DAVID were used for functional analyses of differentially expressed genes. The TRANSFAC database was used to identify oestrogen and progesterone response elements upstream of differentially expressed genes. Cluster analysis demonstrated that natural cycles are associated with a better endometrial receptivity transcriptome than artificial cycles. Artificial cycles seemed to have a stronger negative effect on expression of genes and pathways crucial for endometrial receptivity, including ESR2, FSHR, LEP, and several interleukins and matrix metalloproteinases. Significant overrepresentation of oestrogen response elements among the genes with deteriorated expression in artificial cycles (P < 0.001) was found; progesterone response elements predominated in genes with amended expression with artificial cycles (P = 0.0052).

Diversity and Stability of Lactic Acid Bacteria in Rye Sourdoughs of Four Bakeries with Different Propagation Parameters.

We identified the lactic acid bacteria within rye sourdoughs and starters from four bakeries with different propagation parameters and tracked their dynamics for between 5-28 months after renewal. Evaluation of bacterial communities was performed using plating, denaturing gradient gel electrophoresis, and pyrosequencing of 16S rRNA gene amplicons. Lactobacillus amylovorus and Lactobacillus frumenti or Lactobacillus helveticus, Lactobacillus pontis and Lactobacillus panis prevailed in sourdoughs propagated at higher temperature, while ambient temperature combined with a short fermentation cycle selected for Lactobacillus sanfranciscensis, Lactobacillus pontis, and Lactobacillus zymae or Lactobacillus helveticus, Lactobacillus pontis and Lactobacillus zymae. The ratio of species in bakeries employing room-temperature propagation displayed a seasonal dependence. Introduction of different and controlled propagation parameters at one bakery (higher fermentation temperature, reduced inoculum size, and extended fermentation time) resulted in stabilization of the microbial community with an increased proportion of L. helveticus and L. pontis. Despite these new propagation parameters no new species were detected.

Differences in Gut Microbiota Between Atopic and Healthy Children.

Although gut microbiota has been studied relatively extensively in the context of allergic diseases, there have been several contradictions between these studies. By applying high-throughput sequencing, we aimed to analyze the differences in gut microbiota between atopic and healthy children at 5 and 12 years of age. 51 stool samples were collected from 14 atopic and 15 healthy children and analyzed with 454 pyrosequencing of the 16S rRNA gene. At the ages of 5 and 12 years, Bacteroides, Prevotella, and Dialister dominated gut microbiota in both atopic and healthy groups of children. Children in the atopic group had lower abundance and prevalence of Akkermansia in gut microbiota than their healthy counterparts. Thus, the composition of gut microbiota does not seem to be significantly different between atopic and healthy children, but lower abundance and prevalence of Akkermansia indicate that this bacterium may accompany or play a role in IgE-mediated atopic diseases.

Redox-specialized bacterioplankton metacommunity in a temperate estuary.

This study explored the spatiotemporal dynamics of the bacterioplankton community composition in the Gulf of Finland (easternmost sub-basin of the Baltic Sea) based on phylogenetic analysis of 16S rDNA sequences acquired from community samples via pyrosequencing. Investigations of bacterioplankton in hydrographically complex systems provide good insight into the strategies by which microbes deal with spatiotemporal hydrographic gradients, as demonstrated by our research. Many ribotypes were closely affiliated with sequences isolated from environments with similar steep physiochemical gradients and/or seasonal changes, including seasonally anoxic estuaries. Hence, one of the main conclusions of this study is that marine ecosystems where oxygen and salinity gradients co-occur can be considered a habitat for a cosmopolitan metacommunity consisting of specialized groups occupying niches universal to such environments throughout the world. These niches revolve around functional capabilities to utilize different electron receptors and donors (including trace metal and single carbon compounds). On the other hand, temporal shifts in the bacterioplankton community composition at the surface layer were mainly connected to the seasonal succession of phytoplankton and the inflow of freshwater species. We also conclude that many relatively abundant populations are indigenous and well-established in the area.

High-throughput sequencing approach uncovers the miRNome of peritoneal endometriotic lesions and adjacent healthy tissues.

Accumulating data have shown the involvement of microRNAs (miRNAs) in endometriosis pathogenesis. In this study, we used a novel approach to determine the endometriotic lesion-specific miRNAs by high-throughput small RNA sequencing of paired samples of peritoneal endometriotic lesions and matched healthy surrounding tissues together with eutopic endometria of the same patients. We found five miRNAs specific to epithelial cells--miR-34c, miR-449a, miR-200a, miR-200b and miR-141 showing significantly higher expression in peritoneal endometriotic lesions compared to healthy peritoneal tissues. We also determined the expression levels of miR-200 family target genes E-cadherin, ZEB1 and ZEB2 and found that the expression level of E-cadherin was significantly higher in endometriotic lesions compared to healthy tissues. Further evaluation verified that studied miRNAs could be used as diagnostic markers for confirming the presence of endometrial cells in endometriotic lesion biopsy samples. Furthermore, we demonstrated that the miRNA profile of peritoneal endometriotic lesion biopsies is largely masked by the surrounding peritoneal tissue, challenging the discovery of an accurate lesion-specific miRNA profile. Taken together, our findings indicate that only particular miRNAs with a significantly higher expression in endometriotic cells can be detected from lesion biopsies, and can serve as diagnostic markers for endometriosis.