Description du poste
No treatment has been currently licensed for primary Sjögren's syndrome (pSS), mainly because several factors hamper the drug development for this systemic multi-organ auto-immune disease. In particular, a considerable disease heterogeneity exists among individuals both in terms of clinical manifestations and underpinning biological disturbances. A molecular classification built on multi-omics characterization  has been proposed and identify 4 groups of patients with distinct patterns of immune dysregulation.
To determine a robust signature to stratify patients in different clusters, the most common analytical tools are prediction models like penalized regression, classification such as a support vector machine, decision trees and random forest classifiers or simple differential expression based analysis. These methods do not consider any biological information in the gene selection procedure.
Biological networks, constructed on the basis of the known functions and interactions of individual molecules, offer a rich source of information for driver gene selection. In this internship we aim at investigating methods that use biological a priori information from networks in the selection of most relevant variables as an alternative to classical learning approaches. Such methods have been proposed in [2,3,4] and have the advantage of combining condition specific transcriptome data and allow understanding of the functional role of the individual genes capable of discriminating disease from healthy or between different disease stages
 A new molecular classification to drive precision treatment strategies in primary Sjögren's syndrome Precision medicine approaches, Soret et al, In submission
 Metri, R., Mohan, A., Nsengimana, J. et al. Identification of a gene signature for discriminating metastatic from primary melanoma using a molecular interaction network approach. Sci Rep 7, 17314 (2017). https://doi.org/10.1038/s41598-017-17330-0
 Zhang, W., Chien, J., Yong, J. et al. Network-based machine learning and graph theory algorithms for precision oncology. npj Precision Onc 1, 25 (2017). https://doi.org/10.1038/s41698-017-0029-7
 Lyu Y, Xue L, Zhang F, et al. Condition-adaptive fused graphical lasso (CFGL): An adaptive procedure for inferring condition-specific gene co-expression network. PLoS Comput Biol. 2018;14(9):e1006436. Published 2018 Sep 21. doi:10.1371/journal.pcbi.1006436
Internship will take place in Biomarkers Team from Translationnal medicine department.