CHICAGO – Informatics researchers from the University of Texas Health Science Center (UTHealth) have advanced a computing method to integrate genome-wide association studies with DNA methylation and expression data to improve identification of mechanisms in multiple sclerosis in search of new drug target genes. The work has already led to insights about the potential repurposing of existing drugs to treat MS.
"Because MS is a very complex disease, it has both genetic and environmental factors, potentially 50/50, so integrating genomic data with DNA methylation data and expression data could help us to understand the disease mechanisms in MS," said Astrid Manuel, a PhD student and graduate research assistant in biomedical informatics at UTHealth in Houston.
Manuel presented the updated methodology last month during the online American Medical Informatics Association (AMIA) Informatics Summit. Funded by a grant from the US National Library of Medicine, this research builds on work published a year ago in BMC Medical Genomics, in which Manuel and colleagues only examined GWAS data.
In an interview with GenomeWeb, Manuel noted that previous GWAS and other genetic studies have identified more than 200 variants with "genome-wide significance" in the development and progression of MS. There also has been plenty of methylation-focused epigenetic research that has revealed environmental influences on this disease of the central nervous system.
"Although there is a strong genetic component, there's also an unknown environmental factor," she explained.
Manuel and colleagues applied their computational method as a way of unraveling a bit of the mystery.
According to Manuel, network-based computational analyses are based on a mathematical principle known as graph theory.
In biomedical informatics, this type of analysis features "nodes," representing biological components like genes or proteins, and "edges," symbolizing interactions between the nodes. This method helps uncover insights about the mechanisms at work in disease development and progression.
Her earlier research, originally presented at the 2019 International Conference on Intelligent Biology and Medicine, relied on an informatics tool called Dense Module Search of GWAS (dmGWAS), which was largely developed by Peilin Jia, one of Manuel's professors and coauthors.
The work presented at the AMIA Informatics Summit is a new implementation called Edge-Weighted Dense Module Search of GWAS (EW_dmGWAS), which allowed the UTHealth researchers to integrate GWAS for MS with methylation and expression data in normal-appearing white matter (NAWM) in the brains of MS patients. They built a reference network based on the human protein interactome.
Manuel said that EW_dmGWAS makes "major improvements" in analytics methodology over its predecessor.
This integration required assessment of CpG-level methylation data, which the new algorithm converted to gene-level methylation scores. It then calculated differential coexpression in normal-appearing white matter in MS patients.
"I think that putting together both of the different datasets and integrative methods like network-based analyses can help to get a bigger picture of what's going on," Manuel said.
"Maybe the genetics can give us clues. The epigenetics also gives us clues to the environmental factors that could be potentially an infection because the antigen reaction is not known in MS," Manuel added, noting that some have hypothesized that the disease might be triggered by bacteria, a virus, or other environmental factors.
Denis Bauer, bioinformatics group leader at Australia's Commonwealth Scientific and Industrial Research Organisation (CSIRO), expressed skepticism about of the findings.
"While I 100 percent agree with the premise that gene-gene interactions are important, I am not convinced that they've demonstrated that with this [work]," Bauer said via email.
Bauer wondered how much information really needed to come from protein-protein interactions because MS genes and drug targets are chosen from GWAS summary data anyway. She also said that the authors' conclusions may be too strong because the analysis was performed completely in silico.
"All this supports in my mind is that their approach has selected epigenic regulators," Bauer said.
Manuel said that the results unveiled at AMIA do represent early-stage research, since the presentation was based on an abstract, which Manuel is now working on expanding into a full scientific paper that she intends to submit for publication.
The work described at AMIA and in the BMC Medical Genomics paper was based on the review of brain samples from MS patients to identify immune reactions. Since then, Manuel has switched her research to genomic, epigenomic, and expression analysis of peripheral blood to increase the size of the sample pool and to take a closer look at white blood cells in the immune system.
With this new strategy, Manuel and colleague have been able to find drug signatures of medications approved to treat blood cancer and asthma. "We are [now] looking into those kinds of medications for drug repositioning strategies for MS," she said.
Manuel said that she has been inspired by the US Food and Drug Administration's approval in August 2020 of Novartis' B-cell therapy ofatumumab (Kesimpta) for use in relapsing multiple sclerosis. The FDA originally approved the compound in 2009 for treatment of chronic lymphocytic lymphoma.
Manuel and her team at UTHealth are now looking at repurposing zafirlukast, a leukotriene receptor antagonist currently indicated for asthma. "Leukotriene is a strong mediator of immune reactions, so inhibiting leukotriene could potentially be helpful for multiple sclerosis," she explained.
She also has started to incorporate some clinical and even billing claims data into her analysis to locate drugs that have been prescribed off-label for MS patients.
"We hope to find new drug targets for MS and also drug repositioning strategies for MS," Manuel said. "Hopefully with the clinical data, we can find some drugs that have a positive effect on the symptoms of MS, and that can decrease relapses."