Linear programming-based prediction of immune cell composition

Abstract

Immune cell types play critical roles in pathogenesis of loss of tolerance, resulting in autoimmune disorders. Immune cell composition has now been recognised as a clinical tool for monitoring disease progression, as typically inferred from gene expression information of a large prior set of differentiation genes, demarcating specific immune cell types. However, given a small set of gene families, immune cell type inference has not been established. Here we used Linear Programming (LP), by a Simplex method, to infer fractions of immune cells, based on a small set of genes used in cell surface marker experiments. To evaluate the accuracy of our method, we created multiple simulated data sets, and evaluated their performance against Digital Cell Quantisation (DCQ) and ImmQuant. Finally, we applied LP method to real biological data, from multiple Systemic Lupus Erythematosus (SLE) patients, versus healthy controls, to inspect compositional changes of immune cell types, in SLE pathogenesis. © 2020 Inderscience Enterprises Ltd.