RESUMEN
BACKGROUND: Central collection of distributed medical patient data is problematic due to strict privacy regulations. Especially in clinical environments, such as clinical time-to-event studies, large sample sizes are critical but usually not available at a single institution. It has been shown recently that federated learning, combined with privacy-enhancing technologies, is an excellent and privacy-preserving alternative to data sharing. OBJECTIVE: This study aims to develop and validate a privacy-preserving, federated survival support vector machine (SVM) and make it accessible for researchers to perform cross-institutional time-to-event analyses. METHODS: We extended the survival SVM algorithm to be applicable in federated environments. We further implemented it as a FeatureCloud app, enabling it to run in the federated infrastructure provided by the FeatureCloud platform. Finally, we evaluated our algorithm on 3 benchmark data sets, a large sample size synthetic data set, and a real-world microbiome data set and compared the results to the corresponding central method. RESULTS: Our federated survival SVM produces highly similar results to the centralized model on all data sets. The maximal difference between the model weights of the central model and the federated model was only 0.001, and the mean difference over all data sets was 0.0002. We further show that by including more data in the analysis through federated learning, predictions are more accurate even in the presence of site-dependent batch effects. CONCLUSIONS: The federated survival SVM extends the palette of federated time-to-event analysis methods by a robust machine learning approach. To our knowledge, the implemented FeatureCloud app is the first publicly available implementation of a federated survival SVM, is freely accessible for all kinds of researchers, and can be directly used within the FeatureCloud platform.
RESUMEN
The human microbiome has become an area of intense research due to its potential impact on human health. However, the analysis and interpretation of this data have proven to be challenging due to its complexity and high dimensionality. Machine learning (ML) algorithms can process vast amounts of data to uncover informative patterns and relationships within the data, even with limited prior knowledge. Therefore, there has been a rapid growth in the development of software specifically designed for the analysis and interpretation of microbiome data using ML techniques. These software incorporate a wide range of ML algorithms for clustering, classification, regression, or feature selection, to identify microbial patterns and relationships within the data and generate predictive models. This rapid development with a constant need for new developments and integration of new features require efforts into compile, catalog and classify these tools to create infrastructures and services with easy, transparent, and trustable standards. Here we review the state-of-the-art for ML tools applied in human microbiome studies, performed as part of the COST Action ML4Microbiome activities. This scoping review focuses on ML based software and framework resources currently available for the analysis of microbiome data in humans. The aim is to support microbiologists and biomedical scientists to go deeper into specialized resources that integrate ML techniques and facilitate future benchmarking to create standards for the analysis of microbiome data. The software resources are organized based on the type of analysis they were developed for and the ML techniques they implement. A description of each software with examples of usage is provided including comments about pitfalls and lacks in the usage of software based on ML methods in relation to microbiome data that need to be considered by developers and users. This review represents an extensive compilation to date, offering valuable insights and guidance for researchers interested in leveraging ML approaches for microbiome analysis.
RESUMEN
Microbiome data predictive analysis within a machine learning (ML) workflow presents numerous domain-specific challenges involving preprocessing, feature selection, predictive modeling, performance estimation, model interpretation, and the extraction of biological information from the results. To assist decision-making, we offer a set of recommendations on algorithm selection, pipeline creation and evaluation, stemming from the COST Action ML4Microbiome. We compared the suggested approaches on a multi-cohort shotgun metagenomics dataset of colorectal cancer patients, focusing on their performance in disease diagnosis and biomarker discovery. It is demonstrated that the use of compositional transformations and filtering methods as part of data preprocessing does not always improve the predictive performance of a model. In contrast, the multivariate feature selection, such as the Statistically Equivalent Signatures algorithm, was effective in reducing the classification error. When validated on a separate test dataset, this algorithm in combination with random forest modeling, provided the most accurate performance estimates. Lastly, we showed how linear modeling by logistic regression coupled with visualization techniques such as Individual Conditional Expectation (ICE) plots can yield interpretable results and offer biological insights. These findings are significant for clinicians and non-experts alike in translational applications.
RESUMEN
The rapid development of machine learning (ML) techniques has opened up the data-dense field of microbiome research for novel therapeutic, diagnostic, and prognostic applications targeting a wide range of disorders, which could substantially improve healthcare practices in the era of precision medicine. However, several challenges must be addressed to exploit the benefits of ML in this field fully. In particular, there is a need to establish "gold standard" protocols for conducting ML analysis experiments and improve interactions between microbiome researchers and ML experts. The Machine Learning Techniques in Human Microbiome Studies (ML4Microbiome) COST Action CA18131 is a European network established in 2019 to promote collaboration between discovery-oriented microbiome researchers and data-driven ML experts to optimize and standardize ML approaches for microbiome analysis. This perspective paper presents the key achievements of ML4Microbiome, which include identifying predictive and discriminatory 'omics' features, improving repeatability and comparability, developing automation procedures, and defining priority areas for the novel development of ML methods targeting the microbiome. The insights gained from ML4Microbiome will help to maximize the potential of ML in microbiome research and pave the way for new and improved healthcare practices.
RESUMEN
Although metagenomic sequencing is now the preferred technique to study microbiome-host interactions, analyzing and interpreting microbiome sequencing data presents challenges primarily attributed to the statistical specificities of the data (e.g., sparse, over-dispersed, compositional, inter-variable dependency). This mini review explores preprocessing and transformation methods applied in recent human microbiome studies to address microbiome data analysis challenges. Our results indicate a limited adoption of transformation methods targeting the statistical characteristics of microbiome sequencing data. Instead, there is a prevalent usage of relative and normalization-based transformations that do not specifically account for the specific attributes of microbiome data. The information on preprocessing and transformations applied to the data before analysis was incomplete or missing in many publications, leading to reproducibility concerns, comparability issues, and questionable results. We hope this mini review will provide researchers and newcomers to the field of human microbiome research with an up-to-date point of reference for various data transformation tools and assist them in choosing the most suitable transformation method based on their research questions, objectives, and data characteristics.
RESUMEN
Background: Schizophrenia (SCZ) is a heterogeneous neuropsychiatric disorder for which current treatment has insufficient efficacy and severe adverse effects. The modifiable gut microbiome might be a potential target for intervention to improve neurobiological functions through the gut-microbiome-brain axis. Methods: In this case-control study, gut microbiota of 132 patients with SCZ and increased waist circumference were compared with gut microbiota of two age- and sex-matched control groups, composed of 132 healthy individuals and 132 individuals with metabolic syndrome. Shotgun sequencing was used to characterize fecal samples at the taxonomic and functional levels. Cognition of the patients with SCZ was evaluated using the Brief Assessment of Cognition instrument. Results: SCZ gut microbiota differed significantly from those of healthy control subjects and individuals with metabolic syndrome in terms of richness and global composition. SCZ gut microbiota were notably enriched in Flavonifractor plautii, Collinsella aerofaciens, Bilophila wadsworthia, and Sellimonas intestinalis, while depleted in Faecalibacterium prausnitzii, Ruminococcus lactaris, Ruminococcus bicirculans, and Veillonella rogosae. Functional potential of the gut microbiota accounted for 11% of cognition variability. In particular, the bacterial functional module for synthesizing tyrosine, a precursor for dopamine, was in SCZ cases positively associated with cognitive score (ρ = 0.34, q ≤ .1). Conclusions: Overall, this study shows that the gut microbiome of patients with SCZ differs greatly from that of healthy control subjects or individuals with metabolic syndrome. Cognitive function of patients with SCZ is associated with the potential for gut bacterial biosynthesis of tyrosine, a precursor for dopamine, suggesting that gut microbiota might be an intervention target for alleviation of cognitive dysfunction in SCZ.
RESUMEN
OBJECTIVE: Gut microbiome dysbiosis has previously been reported in spondyloarthritis (SpA) patients and could be critically involved in the pathogenesis of this disorder. The objectives of this study were to further characterize the microbiota structure in SpA patients and to investigate the relationship between dysbiosis and disease activity in light of the putative influence of the genetic background. METHODS: Shotgun sequencing was performed on fecal DNA isolated from stool samples from 2 groups of adult volunteers: SpA patients (n = 102) and healthy controls (n = 63). A subset of the healthy controls comprised the age-matched siblings of patients whose HLA-B27 status was known. Changes in gut microbiota composition were assessed based on species diversity, enterotypes, and taxonomic and functional differences. RESULTS: Dysbiosis was confirmed in SpA patients as compared to healthy controls. The restriction of microbiota diversity was detected in patients with the most active disease, and the abundance of several bacterial species was correlated with Bath Ankylosing Spondylitis Disease Activity Index score. Among healthy controls, significant differences in microbiota composition were also detected between the HLA-B27-positive and the HLA-B27-negative siblings of SpA patients. We highlighted a decreased abundance of several species of bacteria in SpA patients, especially those bacteria belonging to the Clostridiales order. Among the few species of bacteria showing increased abundance, Ruminococcus gnavus was one of the top differentiating species. CONCLUSION: These findings reveal that genetic background and level of disease activity are likely to influence the composition of the gut microbiota of patients with SpA. It may be appropriate for further research on chronic arthritis to focus on these key parameters.
Asunto(s)
Microbioma Gastrointestinal , Microbiota , Espondiloartritis , Adulto , Humanos , Microbioma Gastrointestinal/genética , Antígeno HLA-B27/genética , Disbiosis/microbiología , Espondiloartritis/genética , Espondiloartritis/complicacionesRESUMEN
One of the difficulties encountered in the statistical analysis of metaproteomics data is the high proportion of missing values, which are usually treated by imputation. Nevertheless, imputation methods are based on restrictive assumptions regarding missingness mechanisms, namely "at random" or "not at random". To circumvent these limitations in the context of feature selection in a multi-class comparison, we propose a univariate selection method that combines a test of association between missingness and classes, and a test for difference of observed intensities between classes. This approach implicitly handles both missingness mechanisms. We performed a quantitative and qualitative comparison of our procedure with imputation-based feature selection methods on two experimental data sets, as well as simulated data with various scenarios regarding the missingness mechanisms and the nature of the difference of expression (differential intensity or differential presence). Whereas we observed similar performances in terms of prediction on the experimental data set, the feature ranking and selection from various imputation-based methods were strongly divergent. We showed that the combined test reaches a compromise by correlating reasonably with other methods, and remains efficient in all simulated scenarios unlike imputation-based feature selection methods.
Asunto(s)
Proteómica , Proyectos de Investigación , Exactitud de los DatosRESUMEN
Thanks to the latest developments in mass spectrometry, software and standards, metaproteomics is emerging as the vital complement of metagenomics, to make headway in understanding the actual functioning of living and active microbial communities. Modern metaproteomics offers new possibilities in the area of clinical diagnosis. This is illustrated here, for the still highly challenging diagnosis of intestinal bowel diseases (IBDs). Using bottom-up proteomics, we analyzed the gut metaproteomes of the same twenty faecal specimens processed either fresh or after a two-month freezing period. We focused on metaproteomes of microbial cell envelopes since it is an outstanding way of capturing host and host-microbe interaction signals. The protein profiles of pairs of fresh and frozen-thawed samples were closely related, making feasible deferred analysis in a distant diagnosis centre. The taxonomic and functional landscape of microbes in diverse IBD phenotypes-active ulcerative colitis, or active Crohn's disease either with ileo-colonic or exclusive colonic localization-differed from each other and from the controls. Based on their specific peptides, we could identify proteins that were either strictly overrepresented or underrepresented in all samples of one clinical group compared to all samples of another group, paving the road for promising additional diagnostic tool for IBDs.
Asunto(s)
Colitis Ulcerosa , Enfermedad de Crohn , Microbiota , Biomarcadores/metabolismo , Colitis Ulcerosa/diagnóstico , Enfermedad de Crohn/diagnóstico , Heces , HumanosRESUMEN
SCOPE: An imbalance of the gut microbiota ("dysbiosis") is associated with numerous chronic diseases, and its modulation is a promising novel therapeutic approach. Dietary supplementation with soluble fiber is one of several proposed modulation strategies. This study aims at confirming the impact of the resistant dextrin NUTRIOSE (RD), a soluble fiber with demonstrated beneficial health effects, on the gut microbiota of healthy individuals. METHODS AND RESULTS: Fifty healthy women are enrolled and supplemented daily with either RD (n = 24) or a control product (n = 26) during 6 weeks. Characterization of the fecal metagenome with shotgun sequencing reveals that RD intake dramatically increases the abundance of the commensal bacterium Parabacteroides distasonis. Furthermore, presence in metagenomes of accessory genes from P. distasonis, coding for susCD (a starch-binding membrane protein complex) is associated with a greater increase of the species. This suggests that response to RD might be strain-dependent. CONCLUSION: Supplementation with RD can be used to specifically increase P. distasonis in gut microbiota of healthy women. The magnitude of the response may be associated with fiber-metabolizing capabilities of strains carried by subjects. Further research will seek to confirm that P. distasonis directly modulates the clinical effects observed in other studies.
Asunto(s)
Dextrinas , Suplementos Dietéticos , Bacteroidetes , Dextrinas/farmacología , Dieta , Heces/microbiología , Femenino , HumanosRESUMEN
Current studies are shifting from the use of single linear references to representation of multiple genomes organised in pangenome graphs or variation graphs. Meanwhile, in metagenomic samples, resolving strain-level abundances is a major step in microbiome studies, as associations between strain variants and phenotype are of great interest for diagnostic and therapeutic purposes. We developed StrainFLAIR with the aim of showing the feasibility of using variation graphs for indexing highly similar genomic sequences up to the strain level, and for characterizing a set of unknown sequenced genomes by querying this graph. On simulated data composed of mixtures of strains from the same bacterial species Escherichia coli, results show that StrainFLAIR was able to distinguish and estimate the abundances of close strains, as well as to highlight the presence of a new strain close to a referenced one and to estimate its abundance. On a real dataset composed of a mix of several bacterial species and several strains for the same species, results show that in a more complex configuration StrainFLAIR correctly estimates the abundance of each strain. Hence, results demonstrated how graph representation of multiple close genomes can be used as a reference to characterize a sample at the strain level.
RESUMEN
The number of microbiome-related studies has notably increased the availability of data on human microbiome composition and function. These studies provide the essential material to deeply explore host-microbiome associations and their relation to the development and progression of various complex diseases. Improved data-analytical tools are needed to exploit all information from these biological datasets, taking into account the peculiarities of microbiome data, i.e., compositional, heterogeneous and sparse nature of these datasets. The possibility of predicting host-phenotypes based on taxonomy-informed feature selection to establish an association between microbiome and predict disease states is beneficial for personalized medicine. In this regard, machine learning (ML) provides new insights into the development of models that can be used to predict outputs, such as classification and prediction in microbiology, infer host phenotypes to predict diseases and use microbial communities to stratify patients by their characterization of state-specific microbial signatures. Here we review the state-of-the-art ML methods and respective software applied in human microbiome studies, performed as part of the COST Action ML4Microbiome activities. This scoping review focuses on the application of ML in microbiome studies related to association and clinical use for diagnostics, prognostics, and therapeutics. Although the data presented here is more related to the bacterial community, many algorithms could be applied in general, regardless of the feature type. This literature and software review covering this broad topic is aligned with the scoping review methodology. The manual identification of data sources has been complemented with: (1) automated publication search through digital libraries of the three major publishers using natural language processing (NLP) Toolkit, and (2) an automated identification of relevant software repositories on GitHub and ranking of the related research papers relying on learning to rank approach.
RESUMEN
The gut microbiota are increasingly considered as a main partner of human health. Metaproteomics enables us to move from the functional potential revealed by metagenomics to the functions actually operating in the microbiome. However, metaproteome deciphering remains challenging. In particular, confident interpretation of a myriad of MS/MS spectra can only be pursued with smart database searches. Here, we compare the interpretation of MS/MS data sets from 48 individual human gut microbiomes using three interrogation strategies of the dedicated Integrated nonredundant Gene Catalog (IGC 9.9 million genes from 1267 individual fecal samples) together with the Homo sapiens database: the classical single-step interrogation strategy and two iterative strategies (in either two or three steps) aimed at preselecting a reduced-sized, more targeted search space for the final peptide spectrum matching. Both iterative searches outperformed the single-step classical search in terms of the number of peptides and protein clusters identified and the depth of taxonomic and functional knowledge, and this was the most convincing with the three-step approach. However, iterative searches do not help in reducing variability of repeated analyses, which is inherent to the traditional data-dependent acquisition mode, but this variability did not affect the hierarchical relationship between replicates and all other samples.
Asunto(s)
Microbioma Gastrointestinal , Microbiota , Microbioma Gastrointestinal/genética , Humanos , Metagenómica , Proteómica , Espectrometría de Masas en TándemRESUMEN
The number of indications for fecal microbiota transplantation is expected to rise, thus increasing the needs for production of readily available frozen or freeze-dried transplants. Using shotgun metagenomics, we investigated the capacity of two novel human fecal microbiota transplants prepared in maltodextrin-trehalose solutions (abbreviated MD and TR for maltodextrin:trehalose, 3:1, w/w, and trehalose:maltodextrin 3:1, w/w, respectively), to colonize a germ-free born mouse model. Gavage with frozen-thawed MD or TR suspensions gave the taxonomic profiles of mouse feces that best resembled those obtained with the fresh inoculum (Spearman correlations based on relative abundances of metagenomic species around 0.80 and 0.75 for MD and TR respectively), while engraftment capacity of defrosted NaCl transplants most diverged (Spearman correlations around 0.63). Engraftment of members of the family Lachnospiraceae and Ruminoccocaceae was the most challenging in all groups of mice, being improved with MD and TR transplants compared to NaCl, but still lower than with the fresh preparation. Improvement of engraftment of this important group in maintaining health represents a challenge that could benefit from further research on fecal microbiota transplant manufacturing.
Asunto(s)
Trasplante de Microbiota Fecal/métodos , Microbioma Gastrointestinal , Animales , Criopreservación/métodos , Vida Libre de Gérmenes , Humanos , Masculino , Metagenómica/métodos , Ratones , Ratones Endogámicos C57BLRESUMEN
BACKGROUND AND AIMS: Cirrhosis is associated with changes in gut microbiome composition. Although acute-on-chronic liver failure (ACLF) is the most severe clinical stage of cirrhosis, there is lack of information about gut microbiome alterations in ACLF using quantitative metagenomics. We investigated the gut microbiome in patients with cirrhosis encompassing the whole spectrum of disease (compensated, acutely decompensated without ACLF, and ACLF). A group of healthy subjects was used as control subjects. METHODS: Stool samples were collected prospectively in 182 patients with cirrhosis. DNA library construction and sequencing were performed using the Ion Proton Sequencer (ThermoFisher Scientific, Waltham, MA). Microbial genes were grouped into clusters, denoted as metagenomic species. RESULTS: Cirrhosis was associated with a remarkable reduction in gene and metagenomic species richness compared with healthy subjects. This loss of richness correlated with disease stages and was particularly marked in patients with ACLF and persisted after adjustment for antibiotic therapy. ACLF was associated with a significant increase of Enterococcus and Peptostreptococcus sp and a reduction of some autochthonous bacteria. Gut microbiome alterations correlated with model for end-stage liver disease and Child-Pugh scores and organ failure and was associated with some complications, particularly hepatic encephalopathy and infections. Interestingly, gut microbiome predicted 3-month survival with good stable predictors. Functional analysis showed that patients with cirrhosis had enriched pathways related to ethanol production, γ-aminobutyric acid metabolism, and endotoxin biosynthesis, among others. CONCLUSIONS: Cirrhosis is characterized by marked alterations in gut microbiome that parallel disease stages with maximal changes in ACLF. Altered gut microbiome was associated with complications of cirrhosis and survival. Gut microbiome may contribute to disease progression and poor prognosis. These results should be confirmed in future studies.
Asunto(s)
Insuficiencia Hepática Crónica Agudizada/etiología , Insuficiencia Hepática Crónica Agudizada/patología , Microbioma Gastrointestinal/fisiología , Cirrosis Hepática/etiología , Cirrosis Hepática/patología , Insuficiencia Hepática Crónica Agudizada/mortalidad , Anciano , Estudios de Casos y Controles , Femenino , Humanos , Cirrosis Hepática/mortalidad , Masculino , Metagenómica , Persona de Mediana Edad , Pronóstico , Estudios Prospectivos , Tasa de SupervivenciaRESUMEN
Despite the growing importance of prebiotics in nutrition and gastroenterology, their structural variety is currently still very limited. The lack of straightforward procedures to gain new products in sufficient amounts often hampers application testing and further development. Although the enzyme sucrose phosphorylase can be used to produce the rare disaccharide kojibiose (α-1,2-glucobiose) from the bulk sugars sucrose and glucose, the target compound is only a side product that is difficult to isolate. Accordingly, for this biocatalyst to become economically attractive, the formation of other glucobioses should be avoided and therefore we applied semi-rational mutagenesis and low-throughput screening, which resulted in a double mutant (L341I_Q345S) with a selectivity of 95% for kojibiose. That way, an efficient and scalable production process with a yield of 74% could be established, and with a simple yeast treatment and crystallization step over a hundred grams of highly pure kojibiose (>99.5%) was obtained.
Asunto(s)
Carbohidratos/química , Disacáridos/síntesis química , Glucosiltransferasas/química , Disacáridos/química , Disacáridos/metabolismo , Glucosiltransferasas/metabolismo , Estructura Molecular , PrebióticosRESUMEN
In directed evolution experiments, it is at stake to have methods to screen efficiently the mutant libraries. We propose a web-based tool that implements an established in silico method for the rational screening of mutant libraries. The method, known as ProSAR, attempts to link sequence data to activity. The method uses statistical models trained on small experimental datasets provided by the user. These can integrate potential epistatic interactions between mutations and be used in many diverse biological contexts. It drastically improves the search for leading mutants. The tool is freely available to non-commercial users at http://bo-protscience.fr/prosar/.