PathIntegrate: Multivariate modelling approaches for pathway-based multi-omics data integration

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 9, 2024

As terabytes of multi-omics data are being generated, there is an ever-increasing need for methods facilitating the integration and interpretation such data. Current typically output lists, clusters, or subnetworks molecules related to outcome. Even with expert domain knowledge, discerning biological processes involved a time-consuming activity. Here we propose PathIntegrate, method integrating datasets based on pathways, designed exploit knowledge systems thus provide interpretable models studies. PathIntegrate employs single-sample pathway analysis transform from molecular pathway-level, applies predictive single-view multi-view model integrate Model outputs include pathways ranked by their contribution outcome prediction, each omics layer, importance molecule in pathway. Using semi-synthetic demonstrate benefit grouping into detect signals low signal-to-noise scenarios, as well ability precisely identify important at effect sizes. Finally, using COPD COVID-19 showcase how enables convenient complex high-dimensional datasets. The Python package available https://github.com/cwieder/PathIntegrate.

Language: Английский

---

Effects of a second iron dextran injection administered to piglets during lactation on differential gene expression in liver and duodenum at weaning

Journal of Animal Science, Journal Year: 2024, Volume and Issue: 102

Published: Jan. 1, 2024

Abstract Six female littermate piglets were used in an experiment to evaluate the mRNA expression tissues from given one or two 1 mL injections of iron dextran (200 mg Fe/mL). All litter administered first injection < 24 h after birth. On day 7, paired by weight (mean body = 1.72 ± 0.13 kg) and piglet each pair was randomly selected as control (CON) other received a second (+Fe). At weaning on 22, anesthetized, samples liver duodenum taken anesthetized preserved until extraction. differential gene data analyzed with fold change cutoff (FC) |1.2| P 0.05. Pathway analysis conducted Z-score In 435 genes significantly changed FC ≥ duodenum, Claudin 2 inversely affected + Fe. (CLDN1) plays key role cell-to-cell adhesion epithelial cell sheets upregulated (FC 4.48, 0.0423). (CLDN2) is expressed cation leaky epithelia, especially during disease inflammation downregulated −1.41, 0.0097). liver, 362 The most dose 200 Fe hepcidin antimicrobial peptide (HAMP) 40.8. HAMP liver-produced hormone that main circulating regulator absorption distribution across tissues. It also controls major flows into plasma promoting endocytosis degradation ferroportin (SLC4A1). This leads retention Fe-exporting cells decreased flow plasma. Gene related metabolic pathway changes provides evidence for improved feed conversion growth rates preweaning contemporary pigs companion study. there downregulation clusters associated gluconeogenesis (P 0.05). Concurrently, decrease enzymes required urea production These observations suggest may be less need gluconeogenesis, possibly deaminated amino acids. genomic analyses provided empirical linking phenotypic health improvements.

Language: Английский

---

phoenics: Pathways Longitudinal and Differential Analysis in Metabolomics

Published: May 22, 2024

Perform a differential analysis at pathway level based on metabolite quantifications and information composition.The method is Principal Component Analysis step linear mixed model.Automatic query of metabolic pathways also implemented.

Language: Английский

---

The metabolic role of vitamin D in children’s neurodevelopment: a network study

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: June 26, 2023

Autism spectrum disorder (ASD) is a neurodevelopmental with various proposed environmental risk factors and rapidly increasing prevalence. Mounting evidence suggests potential role of vitamin D deficiency in ASD pathogenesis, though the causal mechanisms remain largely unknown. Here we investigate impact on child neurodevelopment through an integrative network approach that combines metabolomic profiles, clinical traits, data from pediatric cohort. Our results show associated changes metabolic networks tryptophan, linoleic, fatty acid metabolism. These correlate distinct ASD-related phenotypes, including delayed communication skills respiratory dysfunctions. Additionally, our analysis kynurenine serotonin sub-pathways may mediate effect early childhood development. Altogether, findings provide metabolome-wide insights into as therapeutic option for other disorders.

Language: Английский

---

Metabolic Profiling

Elsevier eBooks, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

Language: Английский

---

Step-by-Step Guide to Building a Diagnostic Model Using MetaboAnalyst

Published: Jan. 1, 2024

MetaboAnalyst is an online platform for analyzing and interpreting metabolomics data. Its creators, researchers Jianguo Xia David Wishart, have kindly made it available free use worldwide provided constant improvements. The user has at his disposal many algorithms that require in-depth prior knowledge of various machine learning statistical analysis terms. To streamline the daily utilization these tools guide users in developing a diagnostic model using molecular biomarkers, author crafted tutorial. This comprehensive details each step with examples, explanations graph interpretations, commentary on essential concepts needed successful analysis.

Language: Английский

---

Semisynthetic Simulation for Microbiome Data Analysis

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 17, 2024

Abstract High-throughput sequencing data lie at the heart of modern microbiome research. Effective analysis these requires careful preprocessing, modeling, and interpretation to detect subtle signals avoid spurious associations. In this review, we discuss how simulation can serve as a sandbox test candidate approaches, creating setting that mimics real while providing ground truth. This is particularly valuable for power analysis, methods benchmarking, reliability analysis. We explain probability, multivariate regression concepts behind simulators different implementations make trade-offs between generality, faithfulness, controllability. Recognizing all only approximate reality, review evaluate accurately they reflect key properties. also present case studies demonstrating value in differential abundance testing, dimensionality reduction, network integration. Code examples available an online tutorial ( https://go.wisc.edu/8994yz ) be easily adapted new problem settings.

Language: Английский

---

A synthetic data generation pipeline to reproducibly mirror high-resolution multi-variable peptidomics and real-patient clinical data

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 4, 2024

Abstract Generating high quality, real-world clinical and molecular datasets is challenging, costly time intensive. Consequently, such data should be shared with the scientific community, which however carries risk of privacy breaches. The latter limitation hinders community’s ability to freely share access resolution quality data, are essential especially in context personalised medicine. In this study, we present an algorithm based on Gaussian copulas generate synthetic that retain associations within dimensional (peptidomics) datasets. For purpose, 3,881 from 10 cohorts were employed, containing clinical, demographic, (> 21,500 peptide) variables, outcome for individuals a kidney or heart failure event. High developed portray distribution matrix between peptidomics dataset, these distributions, 2,000 patients was developed. Synthetic maintained capacity reproducibly correlate variables. correlation rho-values individual peptides eGFR real-patient highly similar, both at single peptide level (rho = 0.885, p < 2.2e-308) after classification machine learning models -0.394, 5.21e-127; rho real -0.396, 4.64e-67). External validation performed, using independent multi-centric (n 2,964) chronic disease (CKD, defined as 60 mL/min/1.73m²) those normal function (eGFR > 90 mL/min/1.73m²). Similarly, association external significantly reproduced 0.569, 1.8e-218). Subsequent development classifiers by matrices, resulted predictive values (AUC 0.803 0.867 HF CKD, respectively), demonstrating robustness method generation patient data. proposed pipeline represents solution high-dimensional sharing while maintaining confidentiality.

Language: Английский

---

Semisynthetic simulation for microbiome data analysis

Briefings in Bioinformatics, Journal Year: 2024, Volume and Issue: 26(1)

Published: Nov. 22, 2024

Abstract High-throughput sequencing data lie at the heart of modern microbiome research. Effective analysis these requires careful preprocessing, modeling, and interpretation to detect subtle signals avoid spurious associations. In this review, we discuss how simulation can serve as a sandbox test candidate approaches, creating setting that mimics real while providing ground truth. This is particularly valuable for power analysis, methods benchmarking, reliability analysis. We explain probability, multivariate regression concepts behind simulators different implementations make trade-offs between generality, faithfulness, controllability. Recognizing all only approximate reality, review evaluate accurately they reflect key properties. also present case studies demonstrating value in differential abundance testing, dimensionality reduction, network integration. Code examples available an online tutorial (https://go.wisc.edu/8994yz) be easily adapted new problem settings.

Language: Английский

---