Protein levels were measured in cerebrospinal fluid samples (CSF; n = 186) and plasma samples (n = 165) from persons with multiple sclerosis and healthy controls. CSF samples and plasma samples were taken from 92 persons with CIS or RRMS at Linköping University Hospital, Sweden and 51 persons with CIS or RRMS at the Karolinska University Hospital, Sweden. Everyone fulfilled the revised McDonald criteria from 2010 and 2017 for CIS or Multiple sclerosis (MS). Age-matched healthy controls (HC) were recruited from healthy blood donors (23 at the Linköping University hospital and 20 at the Karolinska University Hospital). The concentration of 1463 proteins were measured using the Olink Explore platform which uses Proximity Extension Assay (PEA) technology. The proteins were preselected from four Olink panels: Explore 384 Cardiometabolic, Explore 384 Inflammation, Explore 384 Neurology, and Explore 384 Oncology. The protein concentrations are given as Olink’s relative protein quantification unit on log2 scale: Normalized Protein Expression (NPX). The NPX values were intensity normalized by Olink.

Sensitive and reliable protein biomarkers are needed to predict disease trajectory and personalize treatment strategies for multiple sclerosis (MS). Here, we use the highly sensitive proximity-extension assay combined with next-generation sequencing (Olink Explore) to quantify 1463 proteins in cerebrospinal fluid (CSF) and plasma from 143 people with early-stage MS and 43 healthy controls. With longitudinally followed discovery and replication cohorts, we identify CSF proteins that consistently predicted both short- and long-term disease progression. Lower levels of neurofilament light chain (NfL) in CSF is superior in predicting the absence of disease activity two years after sampling (replication AUC = 0.77) compared to all other tested proteins. Importantly, we also identify a combination of 11 CSF proteins (CXCL13, LTA, FCN2, ICAM3, LY9, SLAMF7, TYMP, CHI3L1, FYB1, TNFRSF1B and NfL) that predict the severity of disability worsening according to the normalized age-related MS severity score (replication AUC = 0.90). The identification of these proteins may help elucidate pathogenetic processes and might aid decisions on treatment strategies for persons with MS.

Network-based disease modules have proven to be a powerful concept for extracting knowledge about disease mechanisms, predicting for example disease risk factors and side effects of treatments. Plenty of tools exist for the purpose of module inference, but less effort has been put on simultaneously utilizing knowledge about regulatory mechanisms for predicting disease module hub regulators.We developed MODalyseR, a novel software for identifying disease module regulators and reducing modules to the most disease-associated genes. This pipeline integrates and extends previously published software packages MODifieR and ComHub and hereby provides a user-friendly network medicine framework combining the concepts of disease modules and hub regulators for precise disease gene identification from transcriptomics data. To demonstrate the usability of the tool, we designed a case study for multiple sclerosis that revealed IKZF1 as a promising hub regulator, which was supported by independent ChIP-seq data.MODalyseR is available as a Docker image at https://hub.docker.com/r/ddeweerd/modalyser with user guide and installation instructions found at https://gustafsson-lab.gitlab.io/MODalyseR/.Supplementary data are available at Bioinformatics Advances online.

In this article we use high-throughput epigenomics, transcriptomics, and proteomics data to construct fine-graded models of the "protein-coding units"gathering all transcript isoforms and chromatin accessibility peaks associated with more than 4000 genes in humans. Each protein-coding unit has the structure of a directed acyclic graph (DAG) and can be represented as a Bayesian network. The factorization of the joint probability distribution induced by the DAGs imposes a number of conditional independence relationships among the variables forming a protein-coding unit, corresponding to the missing edges in the DAGs. We show that a large fraction of these conditional independencies are indeed verified by the data. Factors driving this verification appear to be the structural and functional annotation of the transcript isoforms, as well as a notion of structural balance (or frustration-free) of the corresponding sample correlation graph, which naturally leads to reduction of correlation (and hence to independence) upon conditioning.

BackgroundEnvironmental exposures may alter DNA methylation patterns of T helper cells. As T helper cells are instrumental for allergy development, changes in methylation patterns may constitute a mechanism of action for allergy preventive interventions. While epigenetic effects of separate perinatal probiotic or omega -3 fatty acid supplementation have been studied previously, the combined treatment has not been assessed. We aimed to investigate epigenome-wide DNA methylation patterns from a sub-group of children in an on-going randomised double-blind placebo-controlled allergy prevention trial using pre- and postnatal combined Lactobacillus reuteri and omega -3 fatty acid treatment. To this end,>866000 CpG sites (MethylationEPIC 850K array) in cord blood CD4+ T cells were examined in samples from all four study arms (double-treatment: n=18, single treatments: probiotics n=16, omega -3 n=15, and double placebo: n=14). Statistical and bioinformatic analyses identified treatment-associated differentially methylated CpGs and genes, which were used to identify putatively treatment-induced network modules. Pathway analyses inferred biological relevance, and comparisons were made to an independent allergy data set.ResultsComparing the active treatments to the double placebo group, most differentially methylated CpGs and genes were hypermethylated, possibly suggesting induction of transcriptional inhibition. The double-treated group showed the largest number of differentially methylated CpGs, of which many were unique, suggesting synergy between interventions. Clusters within the double-treated network module consisted of immune-related pathways, including T cell receptor signalling, and antigen processing and presentation, with similar pathways revealed for the single-treatment modules. CpGs derived from differential methylation and network module analyses were enriched in an independent allergy data set, particularly in the double-treatment group, proposing treatment-induced DNA methylation changes as relevant for allergy development.ConclusionPrenatal L. reuteri and/or omega -3 fatty acid treatment results in hypermethylation and affects immune- and allergy-related pathways in neonatal T helper cells, with potentially synergistic effects between the interventions and relevance for allergic disease. Further studies need to address these findings on a transcriptional level, and whether the results associate to allergy development in the children. Understanding the role of DNA methylation in regulating effects of perinatal probiotic and omega -3 interventions may provide essential knowledge in the development of efficacious allergy preventive strategies.Trial registration ClinicalTrials.gov, ClinicalTrials.gov-ID: NCT01542970. Registered 27th of February 2012-Retrospectively registered, https://clinicaltrials.gov/ct2/show/NCT01542970.

MOTIVATION: High correlation in expression between regulatory elements is a persistent obstacle for the reverse-engineering of gene regulatory networks. If two potential regulators have matching expression patterns, it becomes challenging to differentiate between them, thus increasing the risk of false positive identifications.

RESULTS: To allow for gene regulation predictions of high confidence, we propose a novel method, the Linear Profile Likelihood (LiPLike), that assumes a regression model and iteratively searches for interactions that cannot be replaced by a linear combination of other predictors. To compare the performance of LiPLike with other available inference methods, we benchmarked LiPLike using three independent datasets from the Dialogue on Reverse Engineering Assessment and Methods 5 (DREAM5) network inference challenge. We found that LiPLike could be used to stratify predictions of other inference tools, and when applied to the predictions of DREAM5 participants, we observed an average improvement in accuracy of >140% compared to individual methods. Furthermore, LiPLike was able to independently predict networks better than all DREAM5 participants when applied to biological data. When predicting the Escherichia coli network, LiPLike had an accuracy of 0.38 for the top-ranked 100 interactions, whereas the corresponding DREAM5 consensus model yielded an accuracy of 0.11.

AVAILABILITY AND IMPLEMENTATION: We made LiPLike available to the community as a Python toolbox, available at https://gitlab.com/Gustafsson-lab/liplike. We believe that LiPLike will be used for high confidence predictions in studies where individual model interactions are of high importance, and to remove false positive predictions made by other state-of-the-art gene-gene regulation prediction tools.

SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Motivation: Complex diseases are due to the dense interactions of many disease-associated factors that dysregulate genes that in turn form the so-called disease modules, which have shown to be a powerful concept for understanding pathological mechanisms. There exist many disease module inference methods that rely on somewhat different assumptions, but there is still no gold standard or best-performing method. Hence, there is a need for combining these methods to generate robust disease modules. Results: We developed MODule IdentiFIER (MODifieR), an ensemble R package of nine disease module inference methods from transcriptomics networks. MODifieR uses standardized input and output allowing the possibility to combine individual modules generated from these methods into more robust disease-specific modules, contributing to a better understanding of complex diseases.

Gemcitabine/carboplatin chemotherapy commonly induces myelosuppression, including neutropenia, leukopenia, and thrombocytopenia. Predicting patients at risk of these adverse drug reactions (ADRs) and adjusting treatments accordingly is a long-term goal of personalized medicine. This study used whole-genome sequencing (WGS) of blood samples from 96 gemcitabine/carboplatin-treated non-small cell lung cancer (NSCLC) patients and gene network modules for predicting myelosuppression. Association of genetic variants in PLINK found 4594, 5019, and 5066 autosomal SNVs/INDELs with p ≤ 1 × 10−3 for neutropenia, leukopenia, and thrombocytopenia, respectively. Based on the SNVs/INDELs we identified the toxicity module, consisting of 215 unique overlapping genes inferred from MCODE-generated gene network modules of 350, 345, and 313 genes, respectively. These module genes showed enrichment for differentially expressed genes in rat bone marrow, human bone marrow, and human cell lines exposed to carboplatin and gemcitabine (p < 0.05). Then using 80% of the patients as training data, random LASSO reduced the number of SNVs/INDELs in the toxicity module into a feasible prediction model consisting of 62 SNVs/INDELs that accurately predict both the training and the test (remaining 20%) data with high (CTCAE 3–4) and low (CTCAE 0–1) maximal myelosuppressive toxicity completely, with the receiver-operating characteristic (ROC) area under the curve (AUC) of 100%. The present study shows how WGS, gene network modules, and random LASSO can be used to develop a feasible and tested model for predicting myelosuppressive toxicity. Although the proposed model predicts myelosuppression in this study, further evaluation in other studies is required to determine its reproducibility, usability, and clinical effect.

The protection against tuberculosis induced by the Bacille Calmette Guerin (BCG) vaccine is unpredictable. In our previous study, altered DNA methylation pattern in peripheral blood mononuclear cells (PBMCs) in response to BCG was observed in a subgroup of individuals, whose macrophages killed mycobacteria effectively (responders). These macrophages also showed production of Interleukin-1 beta (IL-1 beta) in response to mycobacterial stimuli before vaccination. Here, we hypothesized that the propensity to respond to the BCG vaccine is reflected in the DNA methylome. We mapped the differentially methylated genes (DMGs) in PBMCs isolated from responders/non-responders at the time point before vaccination aiming to identify possible predictors of BCG responsiveness. We identified 43 DMGs and subsequent bioinformatic analyses showed that these were enriched for actin-modulating pathways, predicting differences in phagocytosis. This could be validated by experiments showing that phagocytosis of mycobacteria, which is an event preceding mycobacteria-induced IL-1 beta production, was strongly correlated with the DMG pattern.

The molecular mechanisms of insulin resistance in Type 2 diabetes have been extensively studied in primary human adipocytes, and mathematical modelling has clarified the central role of attenuation of mammalian target of rapamycin (mTOR) complex 1 (mTORC1) activity in the diabetic state. Attenuation of mTORC1 in diabetes quells insulin-signalling network-wide, except for the mTOR in complex 2 (mTORC2)-catalysed phosphorylation of protein kinase B (PKB) at Ser(473) (PKB-S473P), which is increased. This unique increase could potentially be explained by feedback and interbranch cross-talk signals. To examine if such mechanisms operate in adipocytes, we herein analysed data from an unbiased phosphoproteomic screen in 3T3-L1 adipocytes. Using a mathematical modelling approach, we showed that a negative signal from mTORC1-p70 S6 kinase (S6K) to rictor-mTORC2 in combination with a positive signal from PKB to SIN1-mTORC2 are compatible with the experimental data. This combined cross-branch signalling predicted an increased PKB-S473P in response to attenuation of mTORC1 - a distinguishing feature of the insulin resistant state in human adipocytes. This aspect of insulin signalling was then verified for our comprehensive model of insulin signalling in human adipocytes. Introduction of the cross-branch signals was compatible with all data for insulin signalling in human adipocytes, and the resulting model can explain all data network-wide, including the increased PKB-S473P in the diabetic state. Our approach was to first identify potential mechanisms in data from a phosphoproteomic screen in a cell line, and then verify such mechanisms in primary human cells, which demonstrates how an unbiased approach can support a direct knowledge-based study.