Biology subjects

Suhre, K.

Publications and source records attributed to Suhre, K..

At least 19 recordsLinked to original sources

bioRxiv2026

Genomic correlates of hyperthermostability revisited: a large-scale validation across 1,963 microbial genomes

In 2003, Suhre and Claverie demonstrated that the difference between the fraction of charged amino acids and the fraction of polar uncharged amino acids in a proteome (the CvP-bias) was the single genomic feature that most strongly discriminates hyperthermophilic microorganisms from their mesophilic and thermophilic counterparts. The original analysis was based on 71 completely sequenced genomes available at the time. Here, using modern genome databases -- specifically the Bacterial and Viral Bioinformatics Resource Center (BV-BRC) for curated optimal growth temperature (OGT) metadata and the NCBI RefSeq FTP archive for sequence data -- the same analysis is repeated at approximately 28-fold larger scale, covering 1,963 bacterial and archaeal genomes (103 hyperthermophiles, 409 thermophiles, 1,451 mesophiles). The original finding is confirmed with high statistical confidence: CvP-bias is elevated in hyperthermophiles (mean 11.1 {+/-} 3.1%) relative to mesophiles (5.0 {+/-} 2.0%; ANOVA F = 496, p < 10-170), with a large effect size (Cohens d = 2.32) and area under the receiver operating characteristic curve of 0.94 for binary hyper/meso classification. Principal component analysis confirms that the first principal component, explaining 47% of variance, is loaded by CvP-bias as a major contributor, separating hyperthermophiles from other organisms. These results establish that the CvP-bias signal identified in 2003 is not an artifact of small sample size but a genuine, robust property of hyperthermophilic proteomes.

microbiology↗

bioRxiv2025

Theobromine is Associated with Slower Epigenetic Ageing

Theobromine, a commonly consumed dietary alkaloid derived from cocoa, has been linked to extended lifespan in model organisms and to health benefits in humans. We examined associations between circulating theobromine intake, measured using serum metabolomics, and blood-based epigenetic markers of biological ageing in two European human population-based cohorts. Serum theobromine levels were significantly associated with reduced epigenetic age acceleration, as measured by GrimAge (p<2e-7) and DNAmTL (p<0.001) in over 500 individuals from the TwinsUK cohort, and both signals replicated in 1,160 individuals from the KORA cohort (p = 7.2e-08 and p = 0.007, respectively). Sensitivity analyses including covariates of other cocoa and coffee metabolites suggest that the effect is specific to theobromine. Our findings indicate that the reported beneficial links between theobromine intake on health and ageing extend to the molecular epigenetic level in humans.

genetics↗

bioRxiv2025

Genomic Analysis of Date Palm Fruit Size Trait and Identification of Candidate Genes through GWAS

The commercial value of economically significant fruits, including date palm fruit (dates), is influenced by various factors, such as biochemical composition and morphological features like size, shape, and visual appearance, which are key determinants of their quality and market value. Dates are typically consumed at the dry stage (Tamar), during which they exhibit a wide range of physical characteristics, such as color, length, weight, and skin appearance. Understanding the genetic basis of these traits is crucial for improving crop quality and breeding new cultivars. In this study, we integrated a genome dataset from highly diverse date cultivars with phenotypes of dry fruit such as length, width, area, and weight, identifying multiple significant genetic loci (SNPs) associated with these traits. We also identified candidate genes located near the associated SNPs that are involved in biological processes such as cell differentiation, proliferation, growth, and the regulation of signalling pathways for growth regulators like auxin and abscisic acid, as observed in other plants. Gene expression analysis reveals that many of these genes are highly expressed in the early stage of fruit development when the fruit attains its maximum size and weight. These findings will enhance our understanding of genetic determinants of fruit size particularly at the commercially important Tamar stage.

genomics↗

bioRxiv2024

The genetics of fruit skin separation in date palm

The physical appearance of date palm (Phoenix dactylifera) fruit (dates) is important for its market value. Many date-producing countries experience significant financial losses due to the poor appearance of the fruit, skin separation or puffiness being a major reason. Previous research showed evidence linking the skin separation phenotype to environmental conditions. In this study, we show that there is both an environmental and genetic contribution to the fruit skin separation phenotype. We show that beyond environmental factors, genetics is a strong contributor to the most extreme skin separation in some cultivars. A genome-wide association study was conducted using genome data of 199 samples collected from 14 countries that identified nine genetic loci associated with this phenotype and investigated genes in these regions that may contribute to the phenotype overall. Identifying the genetic factors may help better understand the biology and pathways that lead to the environmental effects on skin separation and improve commercial date production. In conclusion, our key finding is that both environmental and genetic factors contribute to skin separation variation, and improvements in environmental factors alone cannot overcome the extreme level of variation observed in some cultivars.

bioinformatics↗

bioRxiv2024

A genome-wide association study of mass spectrometry proteomics using the Seer Proteograph platform

Genome-wide association studies (GWAS) with proteomics are essential tools for drug discovery. To date, most studies have used affinity proteomics platforms, which have limited discovery to protein panels covered by the available affinity binders. Furthermore, it is not clear to which extent protein epitope changing variants interfere with the detection of protein quantitative trait loci (pQTLs). Mass spectrometry-based (MS) proteomics can overcome some of these limitations. Here we report a GWAS using the MS-based Seer ProteographTM platform with blood samples from a discovery cohort of 1,260 American participants and a replication in 325 individuals from Asia, with diverse ethnic backgrounds. We analysed 1,980 proteins quantified in at least 80% of the samples, out of 5,753 proteins quantified across the discovery cohort. We identified 252 and replicated 90 pQTLs, where 30 of the replicated pQTLs have not been reported before. We further investigated 200 of the strongest associated cis-pQTLs previously identified using the SOMAscan and the Olink platforms and found that up to one third of the affinity proteomics pQTLs may be affected by epitope effects, while another third were confirmed by MS proteomics to be consistent with the hypothesis that genetic variants induce changes in protein expression. The present study demonstrates the complementarity of the different proteomics approaches and reports pQTLs not accessible to affinity proteomics, suggesting that many more pQTLs remain to be discovered using MS-based platforms. Graphical AbstractSummarizing the approach taken to identify potential epitope effects. O_FIG O_LINKSMALLFIG WIDTH=166 HEIGHT=200 SRC="FIGDIR/small/596028v1_ufig1.gif" ALT="Figure 1"> View larger version (58K): org.highwire.dtl.DTLVardef@34d11aorg.highwire.dtl.DTLVardef@18c0211org.highwire.dtl.DTLVardef@dbb003org.highwire.dtl.DTLVardef@1009fa7_HPS_FORMAT_FIGEXP M_FIG C_FIG

genomics↗

bioRxiv2024

The Interpretable Multimodal Machine Learning (IMML) framework reveals pathological signatures of distal sensorimotor polyneuropathy

Distal sensorimotor polyneuropathy (DSPN) is a common neurological disorder in elderly adults and people with obesity, prediabetes and diabetes and is associated with high morbidity and premature mortality. DSPN is a multifactorial disease and not fully understood yet. Here, we developed the Interpretable Multimodal Machine Learning (IMML) framework for predicting DSPN prevalence and incidence based on sparse multimodal data. Exploiting IMMLs interpretability further empowered biomarker identification. We leveraged the population-based KORA F4/FF4 cohort including 1,091 participants and their deep multimodal characterisation, i.e. clinical data, genomics, methylomics, transcriptomics, proteomics, inflammatory proteins and metabolomics. Clinical data alone is sufficient to stratify individuals with and without DSPN (AUROC = 0.752), whilst predicting DSPN incidence 6.5{+/-}0.2 years later strongly benefits from clinical data complemented with two or more molecular modalities (improved {Delta}AUROC >0.1, achieved AUROC of 0.714). Important and interpretable features of incident DSPN prediction include up-regulation of proinflammatory cytokines, down-regulation of SUMOylation pathway and essential fatty acids, thus yielding novel insights in the disease pathophysiology. These may become biomarkers for incident DSPN, guide prevention strategies and serve as proof of concept for the utility of IMML in studying complex diseases.

bioinformatics↗

bioRxiv2023

Missing view completion for multi-timepoints omics data via representation disentanglement and temporal knowledge transfer

Longitudinal multi-view omics data offer unique insights into the temporal dynamics of individual-level physiology, which provides opportunities to advance personalized healthcare. However, the common occurrence of incomplete views makes extrapolation tasks difficult, and there is a lack of tailored methods for this critical issue. Here, we introduce LEOPARD, an innovative approach specifically designed to complete missing views in multi-timepoint omics data. By disentangling longitudinal omics data into content and temporal representations, LEOPARD transfers the temporal knowledge to the omics-specific content, thereby completing missing views. The effectiveness of LEOPARD is validated on three benchmark datasets constructed with data from the MGH COVID study and the KORA cohort, spanning periods from 3 days to 14 years. Compared to conventional imputation methods, such as missForest, PMM, GLMM, and cGAN, LEOPARD yields the most robust results across the benchmark datasets. LEOPARD-imputed data also achieve the highest agreement with observed data in our analyses for age-associated metabolites detection, estimated glomerular filtration rate-associated proteins identification, and chronic kidney disease prediction. Our work takes the first step toward a generalized treatment of missing views in longitudinal omics data, enabling comprehensive exploration of temporal dynamics and providing valuable insights into personalized healthcare.

bioinformatics↗

bioRxiv2023

Progesterone induces meiosis through two obligate co-receptors with PLA2 activity

The steroid hormone progesterone (P4) regulates multiple aspects of reproductive and metabolic physiology. Classical P4 signaling operates through nuclear receptors that regulate transcription. In addition, P4 signals through membrane P4 receptors (mPRs) in a rapid nongenomic modality. Despite the established physiological importance of P4 nongenomic signaling, the details of its signal transduction cascade remain elusive. Here, using Xenopus oocyte maturation as a well- established physiological readout of nongenomic P4 signaling, we identify the lipid hydrolase ABHD2 (/{beta} hydrolase domain-containing protein 2) as an essential mPR{beta} co-receptor to trigger meiosis. We show using functional assays coupled to unbiased and targeted cell-based lipidomics that ABHD2 possesses a phospholipase A2 (PLA2) activity that requires mPR{beta}. This PLA2 activity bifurcates P4 signaling by inducing clathrin-dependent endocytosis of mPR{beta}, resulting in the production of lipid messengers that are G-protein coupled receptors agonists. Therefore, P4 drives meiosis by inducing an ABHD2 PLA2 activity that requires both mPR{beta} and ABHD2 as obligate co-receptors. Significance StatementNongenomic progesterone signaling is important for many physiological functions yet the details of its signaling remain elusive. Here we define the early signaling steps downstream of membrane progesterone receptor {beta} (mPR{beta}) during Xenopus oocyte meiosis. We show that progesterone requires two cell membrane receptors to work in unison to signal. The co-receptor complex possesses lipase activity that produces lipid messenger and induces receptor endocytosis to trigger meiosis progression. Our findings have broad physiological implications because nongenomic progesterone signaling operates in many tissues and regulates reproduction and metabolism.

cell biology↗

bioRxiv2023

AutoFocus: A hierarchical framework to explore multi-omic disease associations spanning multiple scales of biomolecular interaction

Recent advances in high-throughput measurement technologies have enabled the analysis of molecular perturbations associated with disease phenotypes at the multi-omic level. Such perturbations can range in scale from fluctuations of individual molecules to entire biological pathways. Data-driven clustering algorithms have long been used to group interactions into interpretable functional modules; however, these modules are typically constrained to a fixed size or statistical cutoff. Furthermore, modules are often analyzed independently of their broader biological context. Consequently, such clustering approaches limit the ability to explore functional module associations with disease phenotypes across multiple scales. Here, we introduce AutoFocus, a data-driven method that hierarchically organizes biomolecules and tests for phenotype enrichment at every level within the hierarchy. As a result, the method allows disease-associated modules to emerge at any scale. We evaluated this approach using two datasets: First, we explored associations of biomolecules from the multi-omic QMDiab dataset (n = 388) with the well-characterized type 2 diabetes phenotype. Secondly, we utilized the ROS/MAP Alzheimers disease dataset (n = 500), consisting of high-throughput measurements of brain tissue to explore modules associated with multiple Alzheimers Disease-related phenotypes. Our method identifies modules that are multi-omic, span multiple pathways, and vary in size. We provide an interactive tool to explore this hierarchy at different levels and probe enriched modules, empowering users to examine the full hierarchy, delve into biomolecular drivers of disease phenotype within a module, and incorporate functional annotations.

bioinformatics↗

bioRxiv2023

The HuMet Repository: Watching human metabolism at work

The human metabolism constantly responds to stimuli such as food intake, fasting, exercise, and stress, triggering adaptive biochemical processes across multiple metabolic pathways. To understand the role of these processes and disruptions thereof in health and disease, detailed documentation of healthy metabolic responses is needed but still scarce on a time-resolved metabolome-wide level. Here, we present the HuMet Repository, a web-based resource for exploring dynamic metabolic responses to six physiological challenges (exercise, 36 h fasting, oral glucose and lipid loads, mixed meal, cold stress) in healthy subjects. For building this resource, we integrated existing and newly derived metabolomics data measured in blood, urine, and breath samples of 15 young healthy men at up to 56 time points during the six highly standardized challenge tests conducted over four days. The data comprise 1.1 million data points acquired on multiple platforms with temporal profiles of 2,656 metabolites from a broad range of biochemical pathways. By embedding the dataset into an interactive web application, we enable users to easily access, search, filter, analyze, and visualize the time-resolved metabolomic readouts and derived results. Users can put metabolites into their larger context by identifying metabolites with similar trajectories or by visualizing metabolites within holistic metabolic networks to pinpoint pathways of interest. In three showcases, we outline the value of the repository for gaining biological insights and generating hypotheses by analyzing the wash-out of dietary markers, the complementarity of metabolomics platforms in dynamic versus cross-sectional data, and similarities and differences in systemic metabolic responses across challenges. With its comprehensive collection of time-resolved metabolomics data, the HuMet Repository, freely accessible at https://humet.org/, is a reference for normal, healthy responses to metabolic challenges in young males. It will enable researchers with and without computational expertise, to flexibly query the data for their own research into the dynamics of human metabolism.

biochemistry↗

bioRxiv2023

Genetic associations with ratios between protein levels detect new pQTLs and reveal protein-protein interactions

Protein quantitative trait loci (pQTLs) are an invaluable source of information for drug target development as they provide genetic evidence to support protein function, suggest relationships between cis- and trans-associated proteins, and link proteins to disease where they collocate with genetic risk loci for clinical endpoints. Using the recently released Olink proteomics data for 1,463 proteins measured in over 54,000 samples of the UK Biobank we identified and replicated 4,248 associations with 2,821 ratios between protein levels (rQTLs) where the strengths of association at known pQTL loci increased by up to several hundred orders of magnitude. We attribute this increase in statistical power (p-gain) to accounting for genetic and non-genetic variance shared by the two proteins in the ratio pair. Protein pairs with a significant p-gain were 7.6-fold enriched in known protein-protein interactions, suggesting that their ratios reflect biological links between the implicated proteins. We then conducted a GWAS on the 2,821 ratios and identified 2,527 novel rQTLs, increasing the number of discovered genetic signals compared to the original protein-only GWAS by 24.7%. At examples we demonstrate that this approach can identify novel loci of clinical relevance, support causal gene identification, and reveal complex networks of interacting proteins. Taken together, our study adds significant value to the genetic insights that can be derived from the UKB proteomics data and motivates the wider use of ratios in large scale GWAS.

genetics↗

bioRxiv2023

Nanoparticle Enrichment Mass-Spectrometry Proteomics Identifies Protein Altering Variants for Precise pQTL Mapping

Genome-wide association studies (GWAS) with proteomics generate hypotheses on protein function and offer genetic evidence for drug target prioritization. Although most protein quantitative loci (pQTLs) have so far been identified by high-throughput affinity proteomics platforms, these methods also have some limitations, such as uncertainty about target identity, non-specific binding of aptamers, and inability to handle epitope-modifying variants that affect affinity binding. Mass spectrometry (MS) proteomics has the potential to overcome these challenges and broaden the scope of pQTL studies. Here, we employ the recently developed MS-based Proteograph workflow (Seer, Inc.) to quantify over 18,000 unique peptides from almost 3,000 proteins in more than 320 blood samples from a multi-ethnic cohort. We implement a bottom-up MS-proteomics approach for the detection and quantification of blood-circulating proteins in the presence of protein altering variants (PAVs). We identify 184 PAVs located in 137 genes that are significantly associated with their corresponding variant peptides in MS data (MS-PAVs). Half of these MS-PAVs (94) overlap with cis-pQTLs previously identified by affinity proteomics pQTL studies, thus confirming the target specificity of the affinity binders. An additional 54 MS-PAVs overlap with trans-pQTLs (and not cis-pQTLs) in affinity proteomics studies, thus identifying the putatively causal cis-encoded protein and providing experimental evidence for its presence in blood. The remaining 36 MS-PAVs have not been previously reported and include proteins that may be inaccessible to affinity proteomics, such as a variant in the incretin pro-peptide (GIP) that associates with type 2 diabetes and cardiovascular disease. Overall, our study introduces a novel approach for analyzing MS-based proteomics data within the GWAS context, provides new insights relevant to genetics-based drug discovery, and highlights the potential of MS-proteomics technologies when applied at population scale. HighlightsO_LIThis is the first pQTL study that uses the Proteograph (Seer Inc.) mass spectrometry-based proteomics workflow. C_LIO_LIWe introduce a novel bottom-up proteomics approach that accounts for protein altering variants in the detection of pQTLs. C_LIO_LIWe confirm the target and potential epitope effects of affinity binders for cis-pQTLs from affinity proteomics studies. C_LIO_LIWe establish putatively causal proteins for known affinity proteomics trans-pQTLs and confirm their presence in blood. C_LIO_LIWe identify novel protein altering variants in proteins of clinical relevance that may not be accessible to affinity proteomics. C_LI Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=94 SRC="FIGDIR/small/537640v1_ufig1.gif" ALT="Figure 1"> View larger version (32K): org.highwire.dtl.DTLVardef@5db14forg.highwire.dtl.DTLVardef@123c1edorg.highwire.dtl.DTLVardef@90ebb6org.highwire.dtl.DTLVardef@6320ff_HPS_FORMAT_FIGEXP M_FIG C_FIG

genomics↗

bioRxiv2022

Genome-wide analysis of dry (Tamar) Date palm fruit color

Date palm (Phoenix dactylifera) fruit are an economically and culturally significant crop in the Middle East and North Africa. There are hundreds of different commercial cultivars producing dates with distinctive shapes, colors, and sizes. Genetic studies of some Date palm traits have been performed, including for date palm sex-determination, sugar content and fresh fruit colour. In this study, we used genome sequences and image data of 199 dry date fruit (Tamar) samples collected from 14 countries to identify genetic loci associated with the color of this fruit stage. Here, we find loci across multiple linkage groups (LG) associated with dry fruit color phenotype. We recover the previously identified VIR genotype associated with fresh fruit yellow or red color and new associations with the lightness and darkness of dry fruit. This study will add resolution to our understanding of the date palm fruit color phenotype especially at the most commercially important tamar stage.

bioinformatics↗

bioRxiv2022

Auto-Immunoproteomics Analysis of COVID-19 ICU Patients Revealed Increased Levels of Autoantibodies Related to Male Reproductive System

The role of autoantibodies in coronavirus disease (COVID-19) complications is not yet fully understood. The current investigation screened two independent cohorts of 97 COVID-19 patients (Discovery (Disc) cohort from Qatar (n = 49) and Replication (Rep) cohort from New York (n = 48)) utilizing high-throughput KoRectly Expressed (KREX) immunome protein-array technology. Autoantibody responses to 57 proteins were significantly altered in the COVID-19 Disc cohort compared to healthy controls (P [≤] 0.05). The Rep cohort had altered autoantibody responses against 26 proteins compared to non-COVID-19 ICU patients that served as controls. Both cohorts showed substantial similarities (r2 = 0.73) and exhibited higher autoantibodies responses to numerous transcription factors, immunomodulatory proteins, and human disease markers. Analysis of the combined cohorts revealed elevated autoantibody responses against SPANXN4, STK25, ATF4, PRKD2, and CHMP3 proteins in COVID-19 patients. KREX analysis of the specific IgG autoantibody responses indicates that the targeted host proteins are supposedly increased in COVID-19 patients. The autoantigen-autoantibody response was cross-validated for SPANXN4 and STK25 proteins using Uniprot BLASTP and sequence alignment tools. SPANXN4 is essential for spermiogenesis and male fertility, which may predict a potential role for this protein in COVID-19 associated male reproductive tract complications and warrants further research. Significance StatementCoronavirus disease (COVID-19), caused by the SARS-CoV-2 virus, has emerged as a global pandemic with a high morbidity rate and multiorgan complications. It is observed that the host immune system contributes to the varied responses to COVID-19 pathogenesis. Autoantibodies, immune system proteins that mistakenly target the bodys own tissue, may underlie some of this variation. We screened total IgG autoantibody responses against 1,318 human proteins in two COVID-19 patient cohorts. We observed several novel markers in COVID-19 patients that are associated with male fertility, such as sperm protein SPANXN4, STK25, and the apoptotic factor ATF4. Particularly, elevated levels of autoantibodies against the testicular tissue-specific protein SPANXN4 offer significant evidence of anticipating the protein role in COVID-19 associated male reproductive complications.

immunology↗

bioRxiv2021

Systems Biology Analysis of Human Genomes Points to Key Pathways Conferring Spina Bifida Risk

Spina bifida (SB) is a debilitating birth defect caused by multiple gene and environment interactions. Though SB shows non-Mendelian inheritance, genetic factors contribute to an estimated 70% of cases. Nevertheless, identifying human mutations conferring SB risk is challenging due to its relative rarity, genetic heterogeneity, incomplete penetrance and environmental influences that hamper GWAS approaches to untargeted discovery. Thus, SB genetic studies may suffer from population substructure and/or selection bias introduced by typical candidate gene searches. We report a population based, ancestry-matched whole-genome sequence analysis of SB genetic predisposition using a systems biology strategy to interrogate 298 case-control subject genomes (149 pairs). Genes that were enriched in likely gene disrupting (LGD), rare protein-coding variants were subjected to machine learning analysis to identify genes in which LGD variants occur with a different frequency in cases vs. controls and so discriminate between these groups. Those genes with high discriminatory potential for SB significantly enriched pathways pertaining to carbon metabolism, inflammation, innate immunity, cytoskeletal regulation and essential transcriptional regulation, indicating their impact on the pathogenesis of human SB. Additionally, interrogation of conserved non-coding sequences identified robust variant enrichment in regulatory regions of several transcription factors critical to embryonic development. This genome-wide perspective offers an effective approach to interrogation of coding and non-coding sequence variant contributions to rare complex genetic disorders.

genomics↗

bioRxiv2021

Metabolic dysregulations of cancer cells with metastatic potential

Metastasis is the primary cause of cancer related deaths due to the limited number of efficient druggable targets. Signatures of dysregulated cancer metabolism could serve as a roadmap for the determination of new treatment strategies, given their vital role in cancer cell responses to multiple challenges, including nutrient and oxygen availability. However, the metabolic signatures of metastatic cells remain vastly elusive. We conducted untargeted metabolic profiling of cells and growth media of five selected triple negative breast cancer cell lines with high metastatic potential (HMP) (MDA-MB-231, MDA-MB-436, MDA-MB-468) and low metastatic potential (LMP) (BT549, HCC1143). We identified 92 metabolites in cells and 22 in growth medium that display significant differences between LMP and HMP. The HMP cell lines had elevated level of molecules involved in glycolysis, TCA cycle and lipid metabolism. We identified metabolic advantages of cell lines with HMP beyond enhanced glycolysis by pinpointing the role of branched chain amino acids (BCAA) catabolism as well as molecules supporting coagulation and platelet activation as important contributors to the metastatic cascade. The landscape of metabolic dysregulations, characterized in our study, could serve in the future as a roadmap for the identification of treatment strategies targeting cancer cells with enhanced metastatic potential.

cancer biology↗

bioRxiv2021

Variational autoencoders learn universal latent representations of metabolomics data

Dimensionality reduction approaches are commonly used for the deconvolution of high-dimensional metabolomics datasets into underlying core metabolic processes. However, current state-of-the-art methods are widely incapable of detecting nonlinearities in metabolomics data. Variational Autoencoders (VAEs) are a deep learning method designed to learn nonlinear latent representations which generalize to unseen data. Here, we trained a VAE on a large-scale metabolomics population cohort of human blood samples consisting of over 4,500 individuals. We analyzed the pathway composition of the latent space using a global feature importance score, which showed that latent dimensions represent distinct cellular processes. To demonstrate model generalizability, we generated latent representations of unseen metabolomics datasets on type 2 diabetes, schizophrenia, and acute myeloid leukemia and found significant correlations with clinical patient groups. Taken together, we demonstrate for the first time that the VAE is a powerful method that learns biologically meaningful, nonlinear, and universal latent representations of metabolomics data.

bioinformatics↗

bioRxiv2019

A strategy to incorporate prior knowledge into correlation network cutoff selection

Correlation networks are commonly used to statistically extract biological interactions between omics markers. Network edge selection is typically based on the significance of the underlying correlation coefficients. A statistical cutoff, however, is not guaranteed to capture biological reality, and heavily depends on dataset properties such as sample size. We here propose an alternative, innovative approach to address the problem of network reconstruction. Specifically, we developed a cutoff selection algorithm that maximizes the agreement to a given ground truth. We first evaluate the approach on IgG glycomics data, for which the biochemical pathway is known and well-characterized. The optimal network outperforms networks obtained with statistical cutoffs and is robust with respect to sample size. Importantly, we can show that even in the case of incomplete or incorrect prior knowledge, the optimal network is close to the true optimum. We then demonstrate the generalizability of the approach on an untargeted metabolomics and a transcriptomics dataset from The Cancer Genome Atlas (TCGA). For the transcriptomics case, we demonstrate that the optimized network is superior to statistical networks in systematically retrieving interactions that were not included in the biological reference used for the optimization. Overall, this paper shows that using prior information for correlation network inference is superior to using regular statistical cutoffs, even if the prior information is incomplete or partially inaccurate.

bioinformatics↗