Search

In the enduring challenge against disease, advancements in medical technology have empowered clinicians with novel diagnostic platforms. Whilst in some cases, a single test may provide a confident diagnosis, often additional tests are required. However, to strike a balance between diagnostic accuracy and cost-effectiveness, one must rigorously construct the clinical pathways. Here, we developed a framework to build multi-platform precision pathways in an automated, unbiased way, recommending the key steps a clinician would take to reach a diagnosis. We achieve this by developing a confidence score, used to simulate a clinical scenario, where at each stage, either a confident diagnosis is made, or another test is performed. Our framework provides a range of tools to interpret, visualize and compare the pathways, improving communication and enabling their evaluation on accuracy and cost, specific to different contexts. This framework will guide the development of novel diagnostic pathways for different diseases, accelerating the implementation of precision medicine into clinical practice.

Microbiome feedbacks are proposed to influence Parkinson’s disease (PD) pathophysiology. A number of studies have evaluated the impact of oral medication on the gut microbiome (GM) in PD. However, the influence of PD device-assisted therapies (DATs) on the GM remains to be investigated. To profile acute gut microbial community alterations in response to PD DAT initiation. Clinical data and stool samples were collected from 21 PD patients initiating either deep brain stimulation (DBS) or levodopa–carbidopa intestinal gel (LCIG) and ten spousal healthy control (HC) subjects. 16S amplicon sequencing of stool DNA enabled comparison of temporal GM stability between groups and with clinical measures, including disease alterations relative to therapy initiation. We assessed GM response to therapy in the PD group by comparing pre-therapy (− 2 and 0 weeks) with post-therapy initiation timepoints (+ 2 and + 4 weeks) and HCs at baseline (0 weeks). Altered GM compositions were noted between the PD and HC groups at various taxonomic levels, including specific differences for DBS (overrepresentation of Clostridium_XlVa, Bilophila, Parabacteroides, Pseudoflavonifractor and underrepresentation of Dorea) and LCIG therapy (overrepresentation of Pseudoflavonifractor, Escherichia/Shigella, and underrepresentation of Gemmiger). Beta diversity changes were also found over the 4 week post-treatment initiation period. We report on initial short-term GM changes in response to the initiation of PD DATs. Prior to the introduction of the DAT, a PD-associated GM was observed. Following initiation of DAT, several DAT-specific changes in GM composition were identified, suggesting DATs can influence the GM in PD.

Introduction People with chronic kidney disease (CKD) experience reduced quality of life (QoL) because of the high symptom and treatment burden. Limited data exist on the factors associated with overall and domain-specific QoL across all CKD stages. Methods Using data from a prospective, multinational study (Australia, New Zealand, Canada, and Spain) in 1696 participants with CKD, we measured overall and domain-specific QoL (pain, self-care, activity, mobility, anxiety/depression) using the EuroQoL, 5 dimension, 3 level. Multivariable linear regression and logistic modeling were used to determine factors associated with overall and domain-specific QoL. Results QoL for patients with CKD stages 3 to 5 (n = 787; mean, 0.81; SD, 0.20) was higher than in patients on dialysis (n = 415; mean, 0.76; SD, 0.24) but lower than in kidney transplant recipients (n = 494; mean, 0.84; SD, 0.21). Factors associated with reduced overall QoL (β [95% confidence intervals]) included being on dialysis (compared with CKD stages 3–5: –0.06 [–0.08 to –0.03]), female sex (–0.03 [–0.05 to –0.006]), lower educational attainment (– 0.04 [–0.06 to –0.02), lacking a partner (–0.04 [–0.06 to –0.02]), having diabetes (–0.05 [–0.07 to –0.02]), history of stroke (–0.09 [–0.13 to –0.05]), cardiovascular disease (–0.06 [–0.08 to –0.03]), and cancer (–0.03 [–0.06 to –0.009]). Pain (43%) and anxiety/depression (30%) were the most commonly affected domains, with dialysis patients reporting decrements in all 5 domains. Predictors for domain-specific QoL included being on dialysis, presence of comorbidities, lower education, female sex, and lack of a partner. Conclusions Being on dialysis, women with CKD, those with multiple comorbidities, lack of a partner, and lower educational attainment were associated with lower QoL across all stages of CKD., Graphical abstract

Background Non-invasive biomarkers may predict adverse events such as acute rejection after kidney transplantation and may be preferable to existing methods because of superior accuracy and convenience. It is uncertain how these biomarkers, often derived from a single study, perform across different cohorts of recipients. Methods Using a cross-validation framework that evaluates the performance of biomarkers, the aim of this study was to devise an integrated gene signature set that predicts acute rejection in kidney transplant recipients. Inclusion criteria were publicly available datasets of gene signatures that reported acute rejection episodes after kidney transplantation. We tested the predictive probability for acute rejection using gene signatures within individual datasets and validated the set using other datasets. Eight eligible studies of 1454 participants, with a total of 512 acute rejections episodes were included. Results All sets of gene signatures had good positive and negative predictive values (79-96%) for acute rejection within their own cohorts, but the predictability reduced to less than 50% when tested in other independent datasets. By integrating signatures sets with high specificity scores across all studies, a set of 150 genes (included CXCL6, CXCL11, OLFM4 and PSG9) which are known to be associated with immune responses, had reasonable predictive values (varied between 69-90%). Conclusions A set of gene signatures for acute rejection derived from a specific cohort of kidney transplant recipients do not appear to provide adequate prediction in an independent cohort of transplant recipients. However, integration of gene signatures sets with high specificity scores may improve the prediction performance of these markers.

Background Nutrigenomics aims at understanding the interaction between nutrition and gene information. Due to the complex interactions of nutrients and genes, their relationship exhibits non-linearity. One of the most effective and efficient methods to explore their relationship is the nutritional geometry framework which fits a response surface for the gene expression over two prespecified nutrition variables. However, when the number of nutrients involved is large, it is challenging to find combinations of informative nutrients with respect to a certain gene and to test whether the relationship is stronger than chance. Methods for identifying informative combinations are essential to understanding the relationship between nutrients and genes. Results We introduce Local Consistency Nutrition to Graphics (LC-N2G), a novel approach for ranking and identifying combinations of nutrients with gene expression. In LC-N2G, we first propose a model-free quantity called Local Consistency statistic to measure whether there is non-random relationship between combinations of nutrients and gene expression measurements based on (1) the similarity between samples in the nutrient space and (2) their difference in gene expression. Then combinations with small LC are selected and a permutation test is performed to evaluate their significance. Finally, the response surfaces are generated for the subset of significant relationships. Evaluation on simulated data and real data shows the LC-N2G can accurately find combinations that are correlated with gene expression. Conclusion The LC-N2G is practically powerful for identifying the informative nutrition variables correlated with gene expression. Therefore, LC-N2G is important in the area of nutrigenomics for understanding the relationship between nutrition and gene expression information.

The remarkable flexibility and adaptability of ensemble methods and deep learning models have led to the proliferation of their application in bioinformatics research. Traditionally, these two machine learning techniques have largely been treated as independent methodologies in bioinformatics applications. However, the recent emergence of ensemble deep learning—wherein the two machine learning techniques are combined to achieve synergistic improvements in model accuracy, stability and reproducibility—has prompted a new wave of research and application. Here, we share recent key developments in ensemble deep learning and look at how their contribution has benefited a wide range of bioinformatics research from basic sequence analysis to systems biology. While the application of ensemble deep learning in bioinformatics is diverse and multifaceted, we identify and discuss the common challenges and opportunities in the context of bioinformatics research. We hope this Review Article will bring together the broader community of machine learning researchers, bioinformaticians and biologists to foster future research and development in ensemble deep learning, and inspire novel bioinformatics applications that are unattainable by traditional methods. Recent developments in machine learning have seen the merging of ensemble and deep learning techniques. The authors review advances in ensemble deep learning methods and their applications in bioinformatics, and discuss the challenges and opportunities going forward.

Background As a rare subtype of prostate carcinoma, basal cell carcinoma (BCC) has not been studied extensively and thus lacks systematic molecular characterization. Methods Here, we applied single-cell genomic amplification and RNA-Seq to a specimen of human prostate BCC (CK34βE12+ /P63+ /PAP- /PSA- ). The mutational landscape was obtained via whole exome sequencing of the amplification mixture of 49 single cells, and the transcriptomes of 69 single cells were also obtained. Results The five putative driver genes mutated in BCC are CASC5, NUTM1, PTPRC, KMT2C, and TBX3, and the top three nucleotide substitutions are C>T, T>C, and C>A, similar to common prostate cancer. The distribution of the variant allele frequency values indicated that these single cells are from the same tumor clone. The 69 single cells were clustered into tumor, stromal, and immune cells based on their global transcriptomic profiles. The tumor cells specifically express basal cell markers like KRT5, KRT14, and KRT23 and epithelial markers EPCAM, CDH1, and CD24. The transcription factor covariance network analysis showed that the BCC tumor cells have distinct regulatory networks. By comparison with current prostate cancer datasets, we found that some of the bulk samples exhibit basal cell signatures. Interestingly, at single-cell resolution the gene expression patterns of prostate BCC tumor cells show uniqueness compared with that of common prostate cancer-derived circulating tumor cells. Conclusions This study, for the first time, discloses the comprehensive mutational and transcriptomic landscapes of prostate BCC, which lays a foundation for the understanding of its tumorigenesis mechanism and provides new insights into prostate cancers in general.

Cell reprogramming offers a potential treatment to many diseases, by regenerating specialized somatic cells. Despite decades of research, discovering the transcription factors that promote cell reprogramming has largely been accomplished through trial and error, a time-consuming and costly method. A computational model for cell reprogramming, however, could guide the hypothesis formulation and experimental validation, to efficiently utilize time and resources. Current methods often cannot account for the heterogeneity observed in cell reprogramming, or they only make short-term predictions, without modelling the entire reprogramming process. Here, we present scREMOTE, a novel computational model for cell reprogramming that leverages single cell multiomics data, enabling a more holistic view of the regulatory mechanisms at cellular resolution. This is achieved by first identifying the regulatory potential of each transcription factor and gene to uncover regulatory relationships, then a regression model is built to estimate the effect of transcription factor perturbations. We show that scREMOTE successfully predicts the long-term effect of overexpressing two key transcription factors in hair follicle development by capturing higher-order gene regulations. Together, this demonstrates that integrating the multimodal processes governing gene regulation creates a more accurate model for cell reprogramming with significant potential to accelerate research in regenerative medicine.

The microbiome plays a fundamental role in human health and diet is one of the strongest modulators of the gut microbiome. However, interactions between microbiota and host health are complex and diverse. Understanding the interplay between diet, the microbiome and health state could enable the design of personalized intervention strategies and improve the health and wellbeing of affected individuals. A common approach to this is to divide the study population into smaller cohorts based on dietary preferences in the hope of identifying specific microbial signatures. However, classification of patients based solely on diet is unlikely to reflect the microbiome-host health relationship or the taxonomic microbiome makeup. To this end, we present a novel approach, the Nutrition-Ecotype Mixture of Experts (NEMoE) model, for establishing associations between gut microbiota and health state that accounts for diet-specific cohort variability using a regularized mixture of experts model framework with an integrated parameter sharing strategy to ensure data driven diet-cohort identification consistency across taxonomic levels. The success of our approach was demonstrated through a series of simulation studies, in which NEMoE showed robustness with regard to parameter selection and varying degrees of data heterogeneity. Further application to real-world microbiome data from a Parkinson’s disease cohort revealed that NEMoE is capable of not only improving predictive performance for Parkinson’s Disease but also for identifying diet-specific microbiome markers of disease. Our results indicate that NEMoE can be used to uncover diet-specific relationships between nutritional-ecotype and patient health and to contextualize precision nutrition for different diseases.

Liquid chromatography-mass spectrometry-based metabolomics studies are increasingly applied to large population cohorts, which run for several weeks or even years in data acquisition. This inevitably introduces unwanted intra- and inter-batch variations over time that can overshadow true biological signals and thus hinder potential biological discoveries. To date, normalisation approaches have struggled to mitigate the variability introduced by technical factors whilst preserving biological variance, especially for protracted acquisitions. Here, we propose a study design framework with an arrangement for embedding biological sample replicates to quantify variance within and between batches and a workflow that uses these replicates to remove unwanted variation in a hierarchical manner (hRUV). We use this design to produce a dataset of more than 1000 human plasma samples run over an extended period of time. We demonstrate significant improvement of hRUV over existing methods in preserving biological signals whilst removing unwanted variation for large scale metabolomics studies. Our tools not only provide a strategy for large scale data normalisation, but also provides guidance on the design strategy for large omics studies., Mass spectrometry-based metabolomics is a powerful method for profiling large clinical cohorts but batch variations can obscure biologically meaningful differences. Here, the authors develop a computational workflow that removes unwanted data variation while preserving biologically relevant information.

Survival analysis is a branch of statistics that deals with both, the tracking of time and of the survival status simultaneously as the dependent response. Current comparisons of survival model performance mostly center on clinical data with classic statistical survival models, with prediction accuracy often serving as the sole metric of model performance. Moreover, survival analysis approaches for censored omics data have not been thoroughly investigated. The common approach is to binarise the survival time and perform a classification analysis.Here, we develop a benchmarking framework, SurvBenchmark, that evaluates a diverse collection of survival models for both clinical and omics datasets. SurvBenchmark not only focuses on classical approaches such as the Cox model, but it also evaluates state-of-art machine learning survival models. All approaches were assessed using multiple performance metrics, these include model predictability, stability, flexibility and computational issues. Our systematic comparison framework with over 320 comparisons (20 methods over 16 datasets) shows that the performances of survival models vary in practice over real-world datasets and over the choice of the evaluation metric. In particular, we highlight that using multiple performance metrics is critical in providing a balanced assessment of various models. The results in our study will provide practical guidelines for translational scientists and clinicians, as well as define possible areas of investigation in both survival technique and benchmarking strategies.Contactjean.yang@sydney.edu.au

High-throughput single cell technologies hold the promise of discovering novel cellular relationships with disease and necessitate the use of effective analytical workflows. When manual gating is used to define cell types, the gating hierarchy can be used to identify cell types whose abundances change relative to a parent population. This strategy allows subtle changes to be observed that could be missed if small subsets were compared to all measured cells. However, typical analyses that employ unsupervised clustering overlook the valuable hierarchical structure present in cell type definitions by exclusively quantifying the proportions of cell type clusters relative to all cells. We present treekoR, a framework that facilitates multiple quantifications and comparisons of cell type proportions. Our results from twelve case studies reinforce the importance of quantifying proportions relative to parent populations in the analyses of cytometry data -- as failing to do so can lead to missing important biological insights.

We introduce a new method of performing high-dimensional discriminant analysis (DA), which we call multiDA. Starting from multiclass diagonal DA classifiers which avoid the problem of high-dimensio...

© 2018 IEEE. Class labels are required for supervised learning but may be corrupted or missing in various applications. In binary classification, for example, when only a subset of positive instances is labeled whereas the remaining are unlabeled, positive-unlabeled (PU) learning is required to model from both positive and unlabeled data. Similarly, when class labels are corrupted by mislabeled instances, methods are needed for learning in the presence of class label noise (LN). Here we propose adaptive sampling (AdaSampling), a framework for both PU learning and learning with class LN. By iteratively estimating the class mislabeling probability with an adaptive sampling procedure, the proposed method progressively reduces the risk of selecting mislabeled instances for model training and subsequently constructs highly generalizable models even when a large proportion of mislabeled instances is present in the data. We demonstrate the utilities of proposed methods using simulation and benchmark data, and compare them to alternative approaches that are commonly used for PU learning and/or learning with LN. We then introduce two novel bioinformatics applications where AdaSampling is used to: 1) identify kinase-substrates from mass spectrometry-based phosphoproteomics data and 2) predict transcription factor target genes by integrating various next-generation sequencing data.