Biology subjects

Magnusson, R.

Publications and source records attributed to Magnusson, R..

11 recordsLinked to original sources

bioRxiv2026

Detecting and Subtyping Ketoacidosis from Metabolomic Patterns in Forensic Casework

Subtyping of ketoacidosis, a metabolic state characterized by blood acidification due to various causes, remains challenging in forensic casework. Postmortem omics samples paired with machine learning offers an independent tool to address this challenge. However, such data, especially related to real forensic cases, are rare. In Sweden, high-resolution mass spectrometry data routinely collected in forensic toxicology, can be leveraged for metabolomic analysis. Here, we integrate postmortem metabolomics and machine learning models to detect and subtype ketoacidosis-related deaths using real forensic cases in Sweden. From femoral blood samples of 109 alcoholic ketoacidosis cases, 220 diabetic ketoacidosis cases, 140 hypothermia cases, and 1,229 controls (hanging cases), we developed and tested three machine learning models, which achieved over 90% accuracy in ketoacidosis detection and over 80% in subtyping. Validation with independent cohorts (21 starvation cases, 29 alcoholic controls, and 40 diabetic controls) confirmed robustness with over 80% of starvation cases classified as ketoacidosis-related. Feature clustering highlighted metabolites such as cortisol to be important for subtyping. In summary, our findings demonstrate that combining machine learning with postmortem metabolomics enables accurate detection and subtyping of ketoacidosis-related deaths, which is useful for forensic casework.

systems biology↗

bioRxiv2025

Predicting gene regulatory network interactions withhigh certainty using the Linear Profile Likelihood -LiPLike

Gene regulatory network inference typically revolves around the use of expression data to predict the most likely underlying network, and these predictions of transcription factor-target regulations have a low accuracy and are full of false positives. This low accuracy is in part due to a high degree of correlation between regulatory transcription factors. However, most gene regulatory network inference methods aim at predicting the maximizing a likelihood function, overlooking the degrees of freedom in almost equally likely alternative solutions. This chapter introduces the LInear Profile LIKE-lihood, LiPLike, which controls for this uncertainty by only selecting network interactions that are uniquely needed to explain the data. If there exists an alternative solution B to an interaction prediction A, LiPLike predicts neither A nor B, whereas state-of-the-art inference methods typically predict either A or B, or both. LiPLike can be used as a standalone inference method, or to stratify the predictions of alternative inference methods into sets of low and high confidence.

bioinformatics↗

bioRxiv2025

The Human Metabolome and Machine Learning Improves Predictions of the Post-Mortem Interval

An accurate prediction of the time since death, known as the post-mortem interval (PMI), remains a critical research question in forensic and police investigations. Current methods, such as rectal temperature or vitreous potassium levels, only provide reliable PMI estimations up to 48-72 hours. In this study, we utilized metabolomic data from femoral whole blood samples of 4,876 individuals with known PMIs ranging from 1 to 67 days. We developed a neural network model that predicted PMI with a mean/median absolute error of 1.45/1.03 days in unseen test cases, outperforming six other machine learning architectures. To further highlight the biological signal, we performed pseudo time-series clustering of metabolic features used by the model, revealing 158 decreasing, 254 increasing, and 398 features with more complex patterns over the pseudo-time scale. Our findings also indicate that metabolomic data from approximately 256 individuals is sufficient to train a machine learning model for PMI prediction, making this approach widely applicable for researchers and forensic institutes worldwide.

systems biology↗

bioRxiv2023

Representational Learning from Healthy Multi-Tissue Human RNA-seq Data such that Latent Space Arithmetics Extracts Disease Modules

1Developing computational analyses of transcriptomic data has dramatically improved our understanding of complex multifactorial diseases. However, such approaches are limited to small sample sets of disease-affected material, thus being sensitive to statistical biases and noise. Here, we ask if a variational autoencoder (VAE) trained on large groups of healthy, human RNA-seq data of multiple tissues can capture the fundamental healthy gene regulation system such that the learned representation generalizes to account for unseen disease changes. To this end, we trained a multi-scale representation to encode cellular processes ranging from cell types to genegene interactions. Importantly, we found that the learned healthy representations could predict unseen gene expression changes from 25 independent disease datasets. We extracted and decoded disease-specific signals from the VAE latent space to dissect this finding. Interestingly, the gene modules corresponding to this signal contained more disease-specific genes than the respective differential expression analysis in 20 of 25 cases. Finally, we matched genes related to the disease signals to known drug targets. We could extract sets of known and potential pharmaceutical candidates from this analysis and demonstrate the utility in three use cases. In summary, our study showcases how data-driven representation learning using a VAE as a foundational model allows an arithmetic deconstruction of the latent space such that biological insights enable the dissection of disease mechanisms and drug targets. Our model is available at https://github.com/ddeweerd/VAE_Transcriptomics/.

systems biology↗

bioRxiv2023

NetAn: A Python Toolbox Leveraging Network Topology for Comprehensive Gene Annotation Enrichments

BackgroundGene annotation enrichment analysis is the gold standard for studying the biological context of a set of genes, but available tools often overlook important network properties of the underlying gene regulatory system. ResultsWe present the NETwork ANnotation enrichment package, NetAn, built in Python, which augments annotation analysis with approaches such as inference of closely related genes to include local neighbors in the analysis, the extraction of separate network sub-clusters, and the following over-representation analyses based on network clustering. By using NetAn, we demonstrate how these approaches enhance the identification of relevant annotations in human gene sets. In a specific case study on Multiple Sclerosis (MS), NetAns approach of incorporating neighboring genes through network-based expansion demonstrates a distinct advantage in identifying immune-related genes critical to MS pathology. Furthermore, we demonstrate the ability of NetAn to stratify MS annotations to also identify relevant neuron-related enrichments. Lastly, we compare NetAn to alternative network-based approaches, and find it to have greater specificity compared to broader approaches like NET-GE. ConclusionsWe present NetAn, a novel network-based approach that can stratify annotation enrichment analyses by integrating gene interactions and network topology, thereby strengthening biological signals through the inclusion of associated genes. This approach allows for enhanced identification of disease-relevant annotations, as demonstrated in the MS case study.

bioinformatics↗

bioRxiv2021

White-box Deep Neural Network Prediction of Genome-Wide Transcriptome Signatures

Prediction algorithms for protein or gene structures, including transcription factor binding from sequence information, have been transformative in understanding gene regulation. Here we ask whether human transcriptomic profiles can be predicted solely from the expression of transcription factors (TFs). To this end, we explore whether a neural network (NN) could predict the transcriptome from TFs. Using at least one hidden layer, we find that the expression of 1,600 TFs can explain >95% of variance in 25,000 genes. Using the light-up technique to inspect the trained NN, we find an overrepresentation of known TF-gene regulations. Furthermore, the learned prediction network has a hierarchical organization. A smaller set of around 125 core TFs could explain close to 80% of the variance. Interestingly, reducing the number of TFs below 500 induces a rapid decline in prediction performance. Next, we evaluated the prediction model using transcriptional data from 22 human diseases. The TFs were sufficient to predict the target genes dysregulation (rho=0.61, P < 10-216). By inspecting the model, key causative TFs could be extracted for subsequent validation using disease-associated genetic variants. In conclusion, we demonstrate the construction of an interpretable neural network predictor. Analysis of the predictors revealed key TFs that were inducing transcriptional changes during disease.

systems biology↗

bioRxiv2020

Progesterone specifically dampens disease-associated TH1- and TH17-related immune responses during T cell activation in vitro

The changes in progesterone (P4) levels during and after pregnancy coincide with the temporary improvement and worsening of several autoimmune diseases like multiple sclerosis (MS) and rheumatoid arthritis (RA). Most likely immune-endocrine interactions play a major role in these pregnancy-induce effects. In this study, we used next generation sequencing to investigate the direct effects of P4 on CD4+ T cell activation, of central importance in pregnancy and disease. We found that P4 had a profound dampening effect on T cell activation, altering the gene and protein expression profile and opposing many of the changes induced during the activation. The transcriptomic changes induced by P4 were significantly enriched for genes associated with diseases known to be modulated during pregnancy such as MS, RA and psoriasis. The TH1-and TH17-associated transcription factors STAT1 and STAT3 were significantly downregulated by P4 and their downstream targets were significantly enriched among the diseases-associated genes. Several of these genes included well-known and disease-relevant cytokines, such as IL-12{beta}, CXCL10 and OSM, that were further validated also at the protein level using proximity extension assay. Our results extend the previous knowledge of P4 as an immune regulatory hormone and supports its importance during pregnancy for regulating potentially detrimental immune responses towards the semi-allogenic fetus. Further, our results also point toward a potential role for P4 in the pregnancy-induced disease immunomodulation, suggestively through dampening of TH1 and TH17-associated immune responses and highlights the need for further studies evaluating P4 as a future treatment option.

immunology↗

bioRxiv2019

ComHub: Community predictions of hubs in gene regulatory networks

SummaryHub transcription factors, regulating many target genes in gene regulatory networks (GRNs), play important roles as disease regulators and potential drug targets. However, while numerous methods have been developed to predict individual regulator-gene interactions from gene expression data, few methods focus on inferring these hubs. We have developed ComHub, a tool to predict hubs in GRNs. ComHub makes a community prediction of hubs by averaging over predictions by a compendium of network inference methods. Benchmarking ComHub to the DREAM5 challenge data and an independent data set of human gene expression, proved a robust performance of ComHub over all data sets. Lastly, we implemented ComHub to work with both predefined networks and to do standard network inference, which we believe will make it generally applicable. AvailabilityCode is available at https://gitlab.com/Gustafsson-lab/comhub Contactmika.gustafsson@liu.se, rasmus.magnusson@liu.se

bioinformatics↗

bioRxiv2019

LiPLike: Towards gene regulatory network predictions of high-certainty

MotivationReverse engineering of gene regulatory networks has for years struggled with high correlation in expression between regulatory elements. If two regulators have matching expression patterns it is impossible to differentiate between the two, and thus false positive identifications are abundant.\n\nResultsTo allow for gene regulation predictions of high confidence, we propose a novel method, 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 previous DREAM5 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 the accuracy to on average be improved >140% compared to individual methods. Furthermore, we observed that LiPLike independently predicted 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.\n\nAvailabilityWe 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 that LiPLike will be used to remove false positive predictions made by other state-of-the-art gene-gene regulation prediction tools.\n\nContactrasmus.magnusson@liu.se, mika.gustafsson@liu.se

bioinformatics↗

bioRxiv2019

On the prediction of protein abundance from RNA

BackgroundProfiling of mRNA expression is an important method to identify biomarkers but complicated by limited correlations between mRNA expression and protein abundance. We hypothesised that these correlations could be improved by mathematical models based on measuring splice variants and time delay in protein translation. MethodsWe characterised time-series of primary human naive CD4+ T cells during early T-helper type 1 differentiation with RNA-sequencing and mass-spectrometry proteomics. We then performed computational time-series analysis in this system and in two other key human and murine immune cell types. Linear mathematical mixed time-delayed splice variant models were used to predict protein abundances, and the models were validated using out-of-sample predictions. Lastly, we re-analysed RNA-Seq datasets to evaluate biomarker discovery in five T-cell associated diseases, validating the findings for multiple sclerosis (MS) and asthma. ResultsThe new models demonstrated median correlations of mRNA-to-protein abundance of 0.79-0.94, significantly out-performing models not including the usage of multiple splice variants and time-delays, as shown in cross-validation tests. Our mathematical models provided more differentially expressed proteins between patients and controls in all five diseases. Moreover, analysis of these proteins in asthma and MS supported their relevance. One marker, sCD27, was clinically validated in MS using two independent cohorts, for treatment response and prognosis. ConclusionOur splice variant and time-delay models substantially improved the prediction of protein abundance from mRNA data in three immune cell-types. The models provided valuable biomarker candidates, which were validated in clinical studies of MS and asthma. We propose that our strategy is generally applicable for biomarker discovery.

bioinformatics↗

bioRxiv2017

LASSIM - a network inference toolbox for genome-wide mechanistic modeling

Recent technological advancements have made time-resolved, quantitative, multi-omics data available for many model systems, which could be integrated for systems pharmacokinetic use. Here, we present large-scale simulation modeling (LASSIM), which is the first general mathematical tool for performing large-scale inference using mechanistically defined ordinary differential equations (ODE) for gene regulatory networks (GRNs). LASSIM integrates structural knowledge about regulatory interactions and non-linear equations with multiple steady states and dynamic response expression datasets. The rationale behind LASSIM is that biological GRNs can be simplified using a limited subset of core genes that are assumed to regulate all other gene transcription events in the network. LASSIM models are built in two steps, where each step can integrate multiple data-types, and the method is implemented as a general-purpose toolbox using the PyGMo Python package to make the most of multicore computers and high performance clusters, and is available at https://gitlab.com/Gustafsson-lab/lassim. As a method, LASSIM first infers a non-linear ODE system of the pre-specified core genes. Second, LASSIM optimizes the parameters that models the regulation of peripheral genes by core-system genes in parallel. We showed the usefulness of this method by applying LASSIM to infer a large-scale nonlinear model of naive Th2 differentiation, made possible by integrating Th2 specific bindings, time-series and six public and six novel siRNA-mediated knock-down experiments. ChIP-seq showed significant overlap for all tested transcription factors. Next, we performed novel time-series measurements of total T-cells during differentiation towards Th2 and verified that our LASSIM model could monitor those data significantly better than comparable models that used the same Th2 bindings. In summary, the LASSIM toolbox opens the door to a new type of model-based data analysis that combines the strengths of reliable mechanistic models with truly systems-level data. We exemplified the advantage by inferring the first mechanistically motivated genome-wide model of the Th2 transcription regulatory system, which plays an important role in the progression of immune related diseases.\n\nAuthor summaryThere are excellent methods to mathematically model time-resolved biological data on a small scale using accurate mechanistic models. Despite the rapidly increasing availability of such data, mechanistic models have not been applied on a genome-wide level due to excessive runtimes and the non-identifiability of model parameters. However, genome-wide, mechanistic models could potentially answer key clinical questions, such as finding the best drug combinations to induce an expression change from a disease to a healthy state.\n\nWe present LASSIM, which is a toolbox built to infer parameters within mechanistic models on a genomic scale. This is made possible due to a property shared across biological systems, namely the existence of a subset of master regulators, here denoted the core system. The introduction of a core system of genes simplifies the inference into small solvable subproblems, and implies that all main regulatory actions on peripheral genes come from a small set of regulator genes. This separation allows substantial parts of computations to be solved in parallel, i.e. permitting the use of a computer cluster, which substantially reduces the time required for the computation to finish.

systems biology↗