Systemic autoimmune diseases are considered to share genetic susceptibility markers and clinicians expect treatments could benefit from a molecular-based reclassification. In that objective, more than 1,000 patients were recruited by the PRECISESADS project to measure their genotypes and their proteins concentrations in blood.

Two approaches were used to reclassify the patients. First, a novel genome-wide summarization method is evaluated and a density-based clustering workflow enables to find core groups and their genetic signatures in the summarized features. Second, a candidate-based approach is performed using Gaussian mixture models and identifies expected profiles and reveals novel insights about subtypes and symptoms shared among diseases. Finally, to increase the quality and robustness of the clustering, sparse coding feature transformation methods are evaluated and compared.

The newly developed methods enabled to find disease relevant clusters using genome-wide markers and enabled a precise description of expected and novel profiles using disease associated markers.