Combining bottom-up and top-down systems biology methods to obtain an integrative, global RA-specific network

MIAGOUX Quentin, De MEZQUITA Dereck, SING Vidisha, CHALABI Smahane, PETIT-TEIXEIRA Elisabeth, NIARAKIS Anna
2020 Zenodo  
Rheumatoid arthritis (RA) is a multifactorial autoimmune disease that causes chronic inflammation of the joints, with an aetiology still unclear. RA is a complex inflammatory disease that involves an interconnected array of genetic, environmental, and epigenetic factors. Recently, systems biology and network-based approaches have been proposed to study such complex diseases. Combining multiple data types allows us to discover new knowledge as integration can help compensate for missing or
more » ... or missing or unreliable information. Moreover, if multiple sources of evidence point to the same outcome then it is less likely to obtain false positives. Among computational approaches, machine learning stands out as a promising field in bioinformatics, as it allows for integration of multi-omic biomedical datasets. In this work, we present our efforts to integrate different layers of biological data in the form of networks in order to obtain a global view of the disease. To do so, we make use of publicly available transcriptomic datasets (peripheral blood) relative to RA and a variety of bioinformatics analyses such as statistical analysis, differential expression analysis, and machine learning network inference (Nicolle et al., 2015) to infer an RA-specific TF co-regulatory network. The TF cooperativity network is subsequently enriched in signalling cascades and upstream regulators using the state of the art RA map based on the functional overlaps between the two networks (Singh et al., 2020). Lastly, a list of RA specific variants available in DisGeNET (Piñero et al., 2020) will be used as an overlap to highlight key genes associated with known disease mutations. We combine three different biological layers (gene expression, signalling cascades, mutations), obtained by bottom-up prior knowledge-based (DisGenNET, RA map) and top-down data-driven (CoRegNet) methods to build an integrative, disease-specific network. The goal behind this endeavor is to unravel mechanisms governing the regulation of key genes identified as mutation carriers i [...]
doi:10.5281/zenodo.4266123 fatcat:7kxywrsrxvfrppdawk3otvdb6m