LuGENE 2.0 ®: Extending the Computational Pipeline of a Novel Diagnostic Technology for Systemic Lupus Erythematosus; Evaluating the effect of Illumina’s Monopoly in the Sequencing Field on Equitable Access to Precision Medicine Therapies

Gundamraj, Rahul, School of Engineering and Applied Science, University of Virginia
Allen, Timothy, EN-Biomed Engr Dept, University of Virginia
Forelle, MC, Engineering and Society, University of Virginia
Owen, Katherine, AMPEL BioSolutions
Grammer, Amrie, AMPEL BioSolutions
Lipsky, Peter, AMPEL BioSolutions

Systemic Lupus Erythematosus (SLE) is an autoimmune disease that has vast heterogeneity among patients, and has varied clinical manifestation. Because of this, standard diagnosis and treatment are difficult to implement for SLE. Precision medicine is an ideal candidate for this disease because of its heterogeneity, but few precision medicine approaches for SLE involve genomic information. In response to this, AMPEL BioSolutions, LLC has developed a real time blood test named LuGENE®, which uses the RNA expression profile of a patient to be able to classify them into 1 of 8 “groups” of SLE patients. These groups represent molecular endotypes, and are decided based on expression of specific gene sets representing immunologically important cell types or processes that have been related to SLE. As new gene sets are being discovered, it is important to re-evaluate this pipeline and assess whether these 8 endotypes still hold, or if a different number of groups now exist. This study incorporated 6 novel gene sets into a pre-existing pool of 32 modules to re-assess unsupervised clustering and supervised classification of SLE patients based on their RNAseq data. Upon re-evaluation of the pipeline, it was found that 6 molecular endotypes best encapsulated the variation of the RNAseq data compared to the previous 8, with the novel gene sets that were included strongly affecting this clustering. Furthermore, Support Vector Machine and Artificial Neural Network models were constructed to test against the pre-existing Random Forest classifier on these 6 clusters, with Support Vector Machine exhibiting the highest accuracy of 98%. The use of RNAseq data to classify patients into molecular endotypes poses a new standard of SLE diagnosis for years to come.

BS (Bachelor of Science)
Lupus, Machine Learning, Precision Medicine, Gene Set Variation Analysis
Sponsoring Agency:
AMPEL BioSolutions
All rights reserved (no additional license for public reuse)
Issued Date: