Integration of gene expression data with interaction and annotation data reveals patterns of connection between primary Sjogren's syndrome associated genes and immune processes

Background: The availability of whole genome sequences has spurred a
revolution in biological analyses facilitating the development of technologies
for the study of genes and proteins on a cell-wide basis.
Consequently, a range of biological data, including metabolic and
signalling pathways, experimental interaction data, and protein annotations,
is available. There is considerable discordance in data from largescale
gene expression studies of primary SS (pSS). Combining these
data with other types of genome-wide information can provide a more
complete view of the cell in order to identify the key genes and biological
processes that are involved in the pathogenesis of pSS.
Methods: In this study, a list of genes, found to be differentially
expressed between pSS patients and healthy controls in four largescale
microarray studies, was derived from the literature. KEGG
metabolic pathway and GO biological process enrichment analyses
were then performed for the individual datasets and the combined
pSS-associated gene list. Using all the available human interaction
data from the BioGRID database, a functional interaction network was
produced in which nodes represented genes or gene products, and
edges represented any type of interaction between the nodes. The
network was then annotated based on the pathway and process
enrichment results. Finally, the network was filtered to identify patterns
of connectivity between pSS-associated genes and the enriched
cellular processes.
Results: Despite the overlap of differentially expressed genes in the
four datasets being relatively low, there is significant overlap between
the GO biological process enrichments of these genes, with the
majority being related to immune processes. Significant enrichment of
immune-related KEGG pathways was also observed for the pSSassociated
gene list. Following network filtering, a cluster of the pSSassociated
genes was identified. All four gene expression datasets
were represented within this connected component. Several clusters
of genes annotated to the processes innate immune response, multiorganism
process, response to virus and response to stress were
observed in the integrated network, and network filtering revealed
patterns of interaction between these clusters and the pSS-associated
genes. In particular, several pSS-associated genes were found to lie
between clusters of genes annotated to the immune/stress responses
and clusters involved in multi-organism processes.
Conclusion: Gene enrichment and network analyses of the pSSassociated
genes suggest that the innate immune responses, multiorganism
processes, and the responses to virus and to stress are likely
to be involved in pSS pathogenesis. Integration of multiple types of
data in this manner can aid in the interpretation of results since
combining diverse data sources reveals global properties not evident
from a single data source. Future studies may benefit from
incorporating additional detailed clinical data during the analysis of
expression data in order to elucidate the relationship between gene
expression and clinical phenotype.
Disclosure statement: The authors have declared no conflicts of
interest.