Highlighting nonlinear patterns in population genetics datasets

An Improved Kernel Entropy Component Analysis for Damage Detection Under Environmental and Operational Variations

Environmental effects often trigger false alarms in vibration-based damage detection methods used for structural health monitoring (SHM). While conventional techniques like Principal Component Analysis (PCA) and cointegration have been somewhat effective in addressing this issue, challenges such as measurement noise, nonlinear behavior, and non-Gaussian data distribution continue to affect their performance. To address these limitations, a novel damage detection method combining Variational Mode Decomposition (VMD) and Dynamic Kernel Entropy Component Analysis (DKECA) is proposed. The proposed method initially uses the VMD technique to remove seasonal patterns and noise from the modal frequencies. Subsequently, a DKECA model is constructed based on a time-delay data matrix, and the principal components that maximize the Rényi entropy in the high-dimensional space are selected. Using these principal components, a damage detector developed from the T2 statistic is used to determine damage indices for SHM. The effectiveness of the proposed method is verified through both a simulated 7-DOF model and real-world data from the Z24 bridge, with comparative studies highlighting its advantages over existing techniques.

Simplicity within biological complexity

Motivation

Heterogeneous, interconnected, systems-level, molecular (multi-omic) data have become increasingly available and key in precision medicine. We need to utilize them to better stratify patients into risk groups, discover new biomarkers and targets, repurpose known and discover new drugs to personalize medical treatment. Existing methodologies are limited and a paradigm shift is needed to achieve quantitative and qualitative breakthroughs.

Results

In this perspective paper, we survey the literature and argue for the development of a comprehensive, general framework for embedding of multi-scale molecular network data that would enable their explainable exploitation in precision medicine in linear time. Network embedding methods (also called graph representation learning) map nodes to points in low-dimensional space, so that proximity in the learned space reflects the network's topology-function relationships. They have recently achieved unprecedented performance on hard problems of utilizing few omic data in various biomedical applications. However, research thus far has been limited to special variants of the problems and data, with the performance depending on the underlying topology-function network biology hypotheses, the biomedical applications, and evaluation metrics. The availability of multi-omic data, modern graph embedding paradigms and compute power call for a creation and training of efficient, explainable and controllable models, having no potentially dangerous, unexpected behaviour, that make a qualitative breakthrough. We propose to develop a general, comprehensive embedding framework for multi-omic network data, from models to efficient and scalable software implementation, and to apply it to biomedical informatics, focusing on precision medicine and personalized drug discovery. It will lead to a paradigm shift in the computational and biomedical understanding of data and diseases that will open up ways to solve some of the major bottlenecks in precision medicine and other domains.

Computing linkage disequilibrium aware genome embeddings using autoencoders

MOTIVATION

The completion of the genome has paved the way for genome-wide association studies (GWAS), which explained certain proportions of heritability. GWAS are not optimally suited to detect non-linear effects in disease risk, possibly hidden in non-additive interactions (epistasis). Alternative methods for epistasis detection using, e.g. deep neural networks (DNNs) are currently under active development. However, DNNs are constrained by finite computational resources, which can be rapidly depleted due to increasing complexity with the sheer size of the genome. Besides, the curse of dimensionality complicates the task of capturing meaningful genetic patterns for DNNs; therefore necessitates dimensionality reduction.

RESULTS

We propose a method to compress single nucleotide polymorphism (SNP) data, while leveraging the linkage disequilibrium (LD) structure and preserving potential epistasis. This method involves clustering correlated SNPs into haplotype blocks and training per-block autoencoders to learn a compressed representation of the block's genetic content. We provide an adjustable autoencoder design to accommodate diverse blocks and bypass extensive hyperparameter tuning. We applied this method to genotyping data from Project MinE, and achieved 99% average test reconstruction accuracy-i.e. minimal information loss-while compressing the input to nearly 10% of the original size. We demonstrate that haplotype-block based autoencoders outperform linear Principal Component Analysis (PCA) by approximately 3% chromosome-wide accuracy of reconstructed variants. To the extent of our knowledge, our approach is the first to simultaneously leverage haplotype structure and DNNs for dimensionality reduction of genetic data.

AVAILABILITY AND IMPLEMENTATION

Data are available for academic use through Project MinE at https://www.projectmine.com/research/data-sharing/, contingent upon terms and requirements specified by the source studies. Code is available at https://github.com/gizem-tas/haploblock-autoencoders.

Integration of pan-cancer multi-omics data for novel mixed subgroup identification using machine learning methods

Cancer is a heterogeneous disease, and patients with tumors from different organs can share similar epigenetic and genetic alterations. Therefore, it is crucial to identify the novel subgroups of patients with similar molecular characteristics. It is possible to propose a better treatment strategy when the heterogeneity of the patient is accounted for during subgroup identification, irrespective of the tissue of origin. This work proposes a machine learning (ML) based pipeline for subgroup identification in pan-cancer. Here, mRNA, miRNA, DNA methylation, and protein expression features from pan-cancer samples were concatenated and non-linearly projected to a lower dimension using an ML algorithm. This data was then clustered to identify multi-omics-based novel subgroups. The clinical characterization of these ML subgroups indicated significant differences in overall survival (OS) and disease-free survival (DFS) (p-value<0.0001). The subgroups formed by the patients from different tumors shared similar molecular alterations in terms of immune microenvironment, mutation profile, and enriched pathways. Further, decision-level and feature-level fused classification models were built to identify the novel subgroups for unseen samples. Additionally, the classification models were used to obtain the class labels for the validation samples, and the molecular characteristics were verified. To summarize, this work identified novel ML subgroups using multi-omics data and showed that the patients with different tumor types could be similar molecularly. We also proposed and validated the classification models for subgroup identification. The proposed classification models can be used to identify the novel multi-omics subgroups, and the molecular characteristics of each subgroup can be used to design appropriate treatment regimen.

Machine learning based combination of multi-omics data for subgroup identification in non-small cell lung cancer

Non-small Cell Lung Cancer (NSCLC) is a heterogeneous disease with a poor prognosis. Identifying novel subtypes in cancer can help classify patients with similar molecular and clinical phenotypes. This work proposes an end-to-end pipeline for subgroup identification in NSCLC. Here, we used a machine learning (ML) based approach to compress the multi-omics NSCLC data to a lower dimensional space. This data is subjected to consensus K-means clustering to identify the five novel clusters (C1-C5). Survival analysis of the resulting clusters revealed a significant difference in the overall survival of clusters (p-value: 0.019). Each cluster was then molecularly characterized to identify specific molecular characteristics. We found that cluster C3 showed minimal genetic aberration with a high prognosis. Next, classification models were developed using data from each omic level to predict the subgroup of unseen patients. Decision‑level fused classification models were then built using these classifiers, which were used to classify unseen patients into five novel clusters. We also showed that the multi-omics-based classification model outperformed single-omic-based models, and the combination of classifiers proved to be a more accurate prediction model than the individual classifiers. In summary, we have used ML models to develop a classification method and identified five novel NSCLC clusters with different genetic and clinical characteristics.

Growth Kinetics Monitoring of Gram-Negative Pathogenic Microbes Using Raman Spectroscopy

Optical density based measurements are routinely performed to monitor the growth of microbes. These measurements solely depend upon the number of cells and do not provide any information about the changes in the biochemical milieu or biological status. An objective information about these parameters is essential for evaluation of novel therapies and for maximizing the metabolite production. In the present study, we have applied Raman spectroscopy to monitor growth kinetics of three different pathogenic Gram-negative microbes Escherichia coli, Pseudomonas aeruginosa, and Acinetobacter baumannii. Spectral measurements were performed under 532 nm excitation with 5 seconds of exposure time. Spectral features suggest temporal changes in the "peptide" and "nucleic acid" content of cells under different growth stages. Using principal component analysis (PCA), successful discrimination between growth phases was also achieved. Overall, the findings are supportive of the prospective adoption of Raman based approaches for monitoring microbial growth.

KLFDAPC: a supervised machine learning approach for spatial genetic structure analysis

Geographic patterns of human genetic variation provide important insights into human evolution and disease. A commonly used tool to detect and describe them is principal component analysis (PCA) or the supervised linear discriminant analysis of principal components (DAPC). However, genetic features produced from both approaches could fail to correctly characterize population structure for complex scenarios involving admixture. In this study, we introduce Kernel Local Fisher Discriminant Analysis of Principal Components (KLFDAPC), a supervised non-linear approach for inferring individual geographic genetic structure that could rectify the limitations of these approaches by preserving the multimodal space of samples. We tested the power of KLFDAPC to infer population structure and to predict individual geographic origin using neural networks. Simulation results showed that KLFDAPC has higher discriminatory power than PCA and DAPC. The application of our method to empirical European and East Asian genome-wide genetic datasets indicated that the first two reduced features of KLFDAPC correctly recapitulated the geography of individuals and significantly improved the accuracy of predicting individual geographic origin when compared to PCA and DAPC. Therefore, KLFDAPC can be useful for geographic ancestry inference, design of genome scans and correction for spatial stratification in GWAS that link genes to adaptation or disease susceptibility.

A deep learning framework for characterization of genotype data

Dimensionality reduction is a data transformation technique widely used in various fields of genomics research. The application of dimensionality reduction to genotype data is known to capture genetic similarity between individuals, and is used for visualization of genetic variation, identification of population structure as well as ancestry mapping. Among frequently used methods are principal component analysis, which is a linear transform that often misses more fine-scale structures, and neighbor-graph based methods which focus on local relationships rather than large-scale patterns. Deep learning models are a type of nonlinear machine learning method in which the features used in data transformation are decided by the model in a data-driven manner, rather than by the researcher, and have been shown to present a promising alternative to traditional statistical methods for various applications in omics research. In this study, we propose a deep learning model based on a convolutional autoencoder architecture for dimensionality reduction of genotype data. Using a highly diverse cohort of human samples, we demonstrate that the model can identify population clusters and provide richer visual information in comparison to principal component analysis, while preserving global geometry to a higher extent than t-SNE and UMAP, yielding results that are comparable to an alternative deep learning approach based on variational autoencoders. We also discuss the use of the methodology for more general characterization of genotype data, showing that it preserves spatial properties in the form of decay of linkage disequilibrium with distance along the genome and demonstrating its use as a genetic clustering method, comparing results to the ADMIXTURE software frequently used in population genetic studies.

Nonlinear machine learning pattern recognition and bacteria-metabolite multilayer network analysis of perturbed gastric microbiome

The stomach is inhabited by diverse microbial communities, co-existing in a dynamic balance. Long-term use of drugs such as proton pump inhibitors (PPIs), or bacterial infection such as Helicobacter pylori, cause significant microbial alterations. Yet, studies revealing how the commensal bacteria re-organize, due to these perturbations of the gastric environment, are in early phase and rely principally on linear techniques for multivariate analysis. Here we disclose the importance of complementing linear dimensionality reduction techniques with nonlinear ones to unveil hidden patterns that remain unseen by linear embedding. Then, we prove the advantages to complete multivariate pattern analysis with differential network analysis, to reveal mechanisms of bacterial network re-organizations which emerge from perturbations induced by a medical treatment (PPIs) or an infectious state (H. pylori). Finally, we show how to build bacteria-metabolite multilayer networks that can deepen our understanding of the metabolite pathways significantly associated to the perturbed microbial communities.

Performance of model-based multifactor dimensionality reduction methods for epistasis detection by controlling population structure

Background

In genome-wide association studies the extent and impact of confounding due to population structure have been well recognized. Inadequate handling of such confounding is likely to lead to spurious associations, hampering replication, and the identification of causal variants. Several strategies have been developed for protecting associations against confounding, the most popular one is based on Principal Component Analysis. In contrast, the extent and impact of confounding due to population structure in gene-gene interaction association epistasis studies are much less investigated and understood. In particular, the role of nonlinear genetic population substructure in epistasis detection is largely under-investigated, especially outside a regression framework.

Methods

To identify causal variants in synergy, to improve interpretability and replicability of epistasis results, we introduce three strategies based on a model-based multifactor dimensionality reduction approach for structured populations, namely MBMDR-PC, MBMDR-PG, and MBMDR-GC.

Results

Simulation results comparing the performance of various approaches show that in the presence of population structure MBMDR-PC and MBMDR-PG consistently better control type I error rate at the nominal level than MBMDR-GC. Moreover, our proposed three methods of population structure correction outperform MDR-SP in terms of statistical power.

Conclusion

We demonstrate through extensive simulation studies the effect of various degrees of genetic population structure and relatedness on epistasis detection and propose appropriate remedial measures based on linear and nonlinear sample genetic similarity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13040-021-00247-w.

Principals about principal components in statistical genetics

Principal components (PCs) are widely used in statistics and refer to a relatively small number of uncorrelated variables derived from an initial pool of variables, while explaining as much of the total variance as possible. Also in statistical genetics, principal component analysis (PCA) is a popular technique. To achieve optimal results, a thorough understanding about the different implementations of PCA is required and their impact on study results, compared to alternative approaches. In this review, we focus on the possibilities, limitations and role of PCs in ancestry prediction, genome-wide association studies, rare variants analyses, imputation strategies, meta-analysis and epistasis detection. We also describe several variations of classic PCA that deserve increased attention in statistical genetics applications.

Multi-omics integration-a comparison of unsupervised clustering methodologies

With the recent developments in the field of multi-omics integration, the interest in factors such as data preprocessing, choice of the integration method and the number of different omics considered had increased. In this work, the impact of these factors is explored when solving the problem of sample classification, by comparing the performances of five unsupervised algorithms: Multiple Canonical Correlation Analysis, Multiple Co-Inertia Analysis, Multiple Factor Analysis, Joint and Individual Variation Explained and Similarity Network Fusion. These methods were applied to three real data sets taken from literature and several ad hoc simulated scenarios to discuss classification performance in different conditions of noise and signal strength across the data types. The impact of experimental design, feature selection and parameter training has been also evaluated to unravel important conditions that can affect the accuracy of the result.

How to increase our belief in discovered statistical interactions via large-scale association studies?

The understanding that differences in biological epistasis may impact disease risk, diagnosis, or disease management stands in wide contrast to the unavailability of widely accepted large-scale epistasis analysis protocols. Several choices in the analysis workflow will impact false-positive and false-negative rates. One of these choices relates to the exploitation of particular modelling or testing strategies. The strengths and limitations of these need to be well understood, as well as the contexts in which these hold. This will contribute to determining the potentially complementary value of epistasis detection workflows and is expected to increase replication success with biological relevance. In this contribution, we take a recently introduced regression-based epistasis detection tool as a leading example to review the key elements that need to be considered to fully appreciate the value of analytical epistasis detection performance assessments. We point out unresolved hurdles and give our perspectives towards overcoming these.

IPCAPS: an R package for iterative pruning to capture population structure

Background

Resolving population genetic structure is challenging, especially when dealing with closely related or geographically confined populations. Although Principal Component Analysis (PCA)-based methods and genomic variation with single nucleotide polymorphisms (SNPs) are widely used to describe shared genetic ancestry, improvements can be made especially when fine-scale population structure is the target.

Results

This work presents an R package called IPCAPS, which uses SNP information for resolving possibly fine-scale population structure. The IPCAPS routines are built on the iterative pruning Principal Component Analysis (ipPCA) framework that systematically assigns individuals to genetically similar subgroups. In each iteration, our tool is able to detect and eliminate outliers, hereby avoiding severe misclassification errors.

Conclusions

IPCAPS supports different measurement scales for variables used to identify substructure. Hence, panels of gene expression and methylation data can be accommodated as well. The tool can also be applied in patient sub-phenotyping contexts. IPCAPS is developed in R and is freely available from http://bio3.giga.ulg.ac.be/ipcaps.

Ceruloplasmin replacement therapy ameliorates neurological symptoms in a preclinical model of aceruloplasminemia

Aceruloplasminemia is a monogenic disease caused by mutations in the ceruloplasmin gene that result in loss of protein ferroxidase activity. Ceruloplasmin plays a role in iron homeostasis, and its activity impairment leads to iron accumulation in liver, pancreas, and brain. Iron deposition promotes diabetes, retinal degeneration, and progressive neurodegeneration. Current therapies mainly based on iron chelation, partially control systemic iron deposition but are ineffective on neurodegeneration. We investigated the potential of ceruloplasmin replacement therapy in reducing the neurological pathology in the ceruloplasmin-knockout (CpKO) mouse model of aceruloplasminemia. CpKO mice were intraperitoneal administered for 2 months with human ceruloplasmin that was able to enter the brain inducing replacement of the protein levels and rescue of ferroxidase activity. Ceruloplasmin-treated mice showed amelioration of motor incoordination that was associated with diminished loss of Purkinje neurons and reduced brain iron deposition, in particular in the choroid plexus. Computational analysis showed that ceruloplasmin-treated CpKO mice share a similar pattern with wild-type animals, highlighting the efficacy of the therapy. These data suggest that enzyme replacement therapy may be a promising strategy for the treatment of aceruloplasminemia.

Geometric characterisation of disease modules

There is an increasing accumulation of evidence supporting the existence of a hyperbolic geometry underlying the network representation of complex systems. In particular, it has been shown that the latent geometry of the human protein network (hPIN) captures biologically relevant information, leading to a meaningful visual representation of protein-protein interactions and translating challenging systems biology problems into measuring distances between proteins. Moreover, proteins can efficiently communicate with each other, without global knowledge of the hPIN structure, via a greedy routing (GR) process in which hyperbolic distances guide biological signals from source to target proteins. It is thanks to this effective information routing throughout the hPIN that the cell operates, communicates with other cells and reacts to environmental changes. As a result, the malfunction of one or a few members of this intricate system can disturb its dynamics and derive in disease phenotypes. In fact, it is known that the proteins associated with a single disease agglomerate non-randomly in the same region of the hPIN, forming one or several connected components known as the disease module (DM). Here, we present a geometric characterisation of DMs. First, we found that DM positions on the two-dimensional hyperbolic plane reflect their fragmentation and functional heterogeneity, rendering an informative picture of the cellular processes that the disease is affecting. Second, we used a distance-based dissimilarity measure to cluster DMs with shared clinical features. Finally, we took advantage of the GR strategy to study how defective proteins affect the transduction of signals throughout the hPIN.

Reward-enhanced encoding improves relearning of forgotten associations

Research on human memory has shown that monetary incentives can enhance hippocampal memory consolidation and thereby protect memory traces from forgetting. However, it is not known whether initial reward may facilitate the recovery of already forgotten memories weeks after learning. Here, we investigated the influence of monetary reward on later relearning. Nineteen healthy human participants learned object-location associations, for half of which we offered money. Six weeks later, most of these associations had been forgotten as measured by a test of declarative memory. Yet, relearning in the absence of any reward was faster for the originally rewarded associations. Thus, associative memories encoded in a state of monetary reward motivation may persist in a latent form despite the failure to retrieve them explicitly. Alternatively, such facilitation could be analogous to the renewal effect observed in animal conditioning, whereby a reward-associated cue can reinstate anticipatory arousal, which would in turn modulate relearning. This finding has important implications for learning and education, suggesting that even when learned information is no longer accessible via explicit retrieval, the enduring effects of a past prospect of reward could facilitate its recovery.

Integrated Computational Analysis of Genes Associated with Human Hereditary Insensitivity to Pain. A Drug Repurposing Perspective

Genes causally involved in human insensitivity to pain provide a unique molecular source of studying the pathophysiology of pain and the development of novel analgesic drugs. The increasing availability of “big data” enables novel research approaches to chronic pain while also requiring novel techniques for data mining and knowledge discovery. We used machine learning to combine the knowledge about n = 20 genes causally involved in human hereditary insensitivity to pain with the knowledge about the functions of thousands of genes. An integrated computational analysis proposed that among the functions of this set of genes, the processes related to nervous system development and to ceramide and sphingosine signaling pathways are particularly important. This is in line with earlier suggestions to use these pathways as therapeutic target in pain. Following identification of the biological processes characterizing hereditary insensitivity to pain, the biological processes were used for a similarity analysis with the functions of n = 4,834 database-queried drugs. Using emergent self-organizing maps, a cluster of n = 22 drugs was identified sharing important functional features with hereditary insensitivity to pain. Several members of this cluster had been implicated in pain in preclinical experiments. Thus, the present concept of machine-learned knowledge discovery for pain research provides biologically plausible results and seems to be suitable for drug discovery by identifying a narrow choice of repurposing candidates, demonstrating that contemporary machine-learned methods offer innovative approaches to knowledge discovery from available evidence.

Clustering: how much bias do we need?

Scientific investigations in medicine and beyond increasingly require observations to be described by more features than can be simultaneously visualized. Simply reducing the dimensionality by projections destroys essential relationships in the data. Similarly, traditional clustering algorithms introduce data bias that prevents detection of natural structures expected from generic nonlinear processes. We examine how these problems can best be addressed, where in particular we focus on two recent clustering approaches, Phenograph and Hebbian learning clustering, applied to synthetic and natural data examples. Our results reveal that already for very basic questions, minimizing clustering bias is essential, but that results can benefit further from biased post-processing.This article is part of the themed issue 'Mathematical methods in medicine: neuroscience, cardiology and pathology'.

Enlightening discriminative network functional modules behind Principal Component Analysis separation in differential-omic science studies

Omic science is rapidly growing and one of the most employed techniques to explore differential patterns in omic datasets is principal component analysis (PCA). However, a method to enlighten the network of omic features that mostly contribute to the sample separation obtained by PCA is missing. An alternative is to build correlation networks between univariately-selected significant omic features, but this neglects the multivariate unsupervised feature compression responsible for the PCA sample segregation. Biologists and medical researchers often prefer effective methods that offer an immediate interpretation to complicated algorithms that in principle promise an improvement but in practice are difficult to be applied and interpreted. Here we present PC-corr: a simple algorithm that associates to any PCA segregation a discriminative network of features. Such network can be inspected in search of functional modules useful in the definition of combinatorial and multiscale biomarkers from multifaceted omic data in systems and precision biomedicine. We offer proofs of PC-corr efficacy on lipidomic, metagenomic, developmental genomic, population genetic, cancer promoteromic and cancer stem-cell mechanomic data. Finally, PC-corr is a general functional network inference approach that can be easily adopted for big data exploration in computer science and analysis of complex systems in physics.

Gender, Contraceptives and Individual Metabolic Predisposition Shape a Healthy Plasma Lipidome

Lipidomics of human blood plasma is an emerging biomarker discovery approach that compares lipid profiles under pathological and physiologically normal conditions, but how a healthy lipidome varies within the population is poorly understood. By quantifying 281 molecular species from 27 major lipid classes in the plasma of 71 healthy young Caucasians whose 35 clinical blood test and anthropometric indices matched the medical norm, we provided a comprehensive, expandable and clinically relevant resource of reference molar concentrations of individual lipids. We established that gender is a major lipidomic factor, whose impact is strongly enhanced by hormonal contraceptives and mediated by sex hormone-binding globulin. In lipidomics epidemiological studies should avoid mixed-gender cohorts and females taking hormonal contraceptives should be considered as a separate sub-cohort. Within a gender-restricted cohort lipidomics revealed a compositional signature that indicates the predisposition towards an early development of metabolic syndrome in ca. 25% of healthy male individuals suggesting a healthy plasma lipidome as resource for early biomarker discovery.

Mining protein interactomes to improve their reliability and support the advancement of network medicine

High-throughput detection of protein interactions has had a major impact in our understanding of the intricate molecular machinery underlying the living cell, and has permitted the construction of very large protein interactomes. The protein networks that are currently available are incomplete and a significant percentage of their interactions are false positives. Fortunately, the structural properties observed in good quality social or technological networks are also present in biological systems. This has encouraged the development of tools, to improve the reliability of protein networks and predict new interactions based merely on the topological characteristics of their components. Since diseases are rarely caused by the malfunction of a single protein, having a more complete and reliable interactome is crucial in order to identify groups of inter-related proteins involved in disease etiology. These system components can then be targeted with minimal collateral damage. In this article, an important number of network mining tools is reviewed, together with resources from which reliable protein interactomes can be constructed. In addition to the review, a few representative examples of how molecular and clinical data can be integrated to deepen our understanding of pathogenesis are discussed.