Updates & Features
Functional Omics Approach to Identify Overlapping and Unique Pain Pathways
June 2019
Achim Kless, John Bothmer
Grünenthal Innovation, Translational Science and Intelligence
The broad diversity of chronic pain conditions requires a better understanding of the underlying pathophysiologies, which can be described through disease pathways.1,2 The aim of this article is to describe how we have generated an integrated disease pathway map for pain that contains the major pathophysiological components, (including involved cell types, miRNA and inflammatory regulations) that allow us to identify overlapping and unique pathways for disease understanding and identification of novel therapeutic interventions.3
To kick off this undertaking, we identified several key questions that have guided us to integrate the available information and build a big pathway map for pain. However, omics data generated from diverse study types need to be annotated and validated. Since most of the published works present a vast number of pathways, it is difficult to extract relevant key pathways for therapeutic intervention out of unstructured data. Therefore, a simple comprehensive review of available literature results is necessary.
The concept of mapping disease pain pathways has led to many questions: are there any common or unique indications in specific pathways? Can we separate pain indications by such pathways on the molecular level? Do key pathways enhance our view of disease understanding regarding involved pathophysiologies? Finally, can we visualise findings in a simple way so that the design of new biomarkers, drug repurposing and novel targets can be derived? In order to start answering these questions, we started to extract and integrate omics data from the literature on eight relevant pain indications that cover a large part of known pathophysiologies (Figure 1).
| • Dorsal horn root ganglion neuron |
| • Ca/Na ion channels, ASICs, P2X, GluR, TRPs, NMDA |
| • miRNA regulation |
| • GABA-, Glutamate-, Histamine- ergic neurons |
| • Microglia/Astrocytes |
| • Inflammation/Mediators |
| • Neuropeptide/hormone signaling |
| • Blood vessels |
| • Apoptosis |
| • Oxidative stress |
| • Primary sensory neuron |
| • Adrenergic pain modulation |
| • Complement system |
Figure 1 – List of components and pathophysiology
As the first step, we utilised generic PubMed search queries for complex regional pain syndrome (CRPS), endometriosis, vulvodynia, spinal cord injury pain, radiculopathy, small fibre neuropathy (including chemotherapy-induced and diabetic peripheral diabetic neuropathies), fibromyalgia and chronic post-operative pain. We have combined these indications with additional keywords linking to experimental omics results, creating a general formula for the queries, shown below alongside an example workflow for CRPS (Figure 2).
[Formula depiction:]
((disease indication) AND ((genomics) OR (genetics) OR (proteomics) OR (transcriptomics) OR (GWAS) OR (omics) OR (metabolomics) OR (gene) OR (SNP) OR (protein) OR (RNA) OR (microarray) OR (association))
Figure 2 – Workflow example for CRPS
The corresponding publications (~1,000) were manually extracted and curated in a form that allows us to assign each entry to a gene (Table 1). These flat database files of regulated genes, miRNA, and proteins are then categorised as either up (1) or downregulated (-1), or unspecified (0) in cases where reported results did not reveal a clear picture (Table 2). In the next step, we built pathways for each indication performing a functional enrichment analysis, also known as over-representation analysis (ORA) or gene set enrichment analysis, on these gene lists. The number of known pathways and contributing genes are compared to the number of genes present in a dataset (Table 3). On the basis of statistics, a ranking of identified pathways can be achieved that can be represented by sets of known gene ontology (GO) terms which, for instance, describe ‘cytochrome c’ by the molecular function oxidoreductase activity, the biological process oxidative phosphorylation, and the cellular component mitochondrial matrix.4–6
Table 1 – Flat database
Table 2 – Categorisation of regulated genes/proteins
Table 3 – The most relevant GO-processes
We repeated the processes for all eight indications and integrated the 20 highest ranking pathways for each indication into a big pain map. An example of such a ranking is given in (Figure 3).4 Since single network nodes in that map can be marked by thermometers from any dataset, it allows direct comparison of novel information on the basis of regulated genes, RNA or proteins through a single view. Go ahead and try for yourself to see what happens when you select several of the indications! What differences do you see between the indications?
Figure 3 – Ranking of pathways
To summarise, PubMed searches were used to compile omics results from genomics, proteomics, transcriptomics and miRNA publications to build databases of associations and regulations in disease tissue. Subsequent pathway analysis using gene set enrichment analysis was then integrated into a big pain map.3 Overlapping and unique pain pathways in eight different indications have been identified, including relevant descriptions of the pathophysiologies: dorsal horn root ganglion neuron, Ca2+/Na+ ion channels, ASICs, P2X, GluR, TRPs, NMDA, miRNA regulation, GABA-, glutamate- and histaminergic neurons, microglia/astrocytes, inflammation/mediators, neuropeptide/hormone signaling, blood vessels, apoptosis, oxidative stress, primary sensory neuron, adrenergic pain modulation and the complement system.
In essence, our conclusion is that it is possible to generate an integrated disease pathway map to elucidate disease understanding and underlying pathophysiology in pain, as demonstrated in our identification of overlapping and unique pathways for eight indications.
Please take a look at the differences of involved pain pathways by selecting the different indications in the Pain Pathways map.
Would you like to see how your data is represented in this pain map? We are seeking partners to further refine our results with different indications and pathophysiologies and would like to collaborate with you! If you would like to see more details and are interested in collaborating with us to analyse your own data set, please contact: achim.kless@grunenthal.com.
In the next part of this series, we will integrate publicly available genome-wide association study (GWAS) data and take a step towards the interpretation of GWAS results compared to information derived from gene–disease associations in smaller cohorts.
References
- EE Young, et al. Genetic basis of pain variability: Recent advances, J Med Genet. 2012; 49(1): 1–9
- K Zorina-Lichtenwalter, CB Meloto, S Khoury, L Diatchenko. Genetic Predictors of human chronic pain conditions. Neuroscience. 2016; 338: 36–62
- A Ultsch, et al. A data science approach to candidate gene selection of pain regarded as a process of learning and neural plasticity. Pain. 2016; 157(12): 2747–57
- Clarivate Analytics. Metacore 2018, https://portal.genego.com/
- Reimand J, et al. g:Profiler -- a web server for functional interpretation of gene lists (2016 update). Nucleic Acids Res. 2016 8; 44(W1):W83–9
- Supek F, et al. REVIGO summarizes and visualizes long lists of Gene Ontology terms. PLoS One. 2011; 6(7): e21800