Integration of single-cell RNA-seq data into population models to characterize cancer metabolism

Tags

Top Highlights

• To fully uncover the metabolic heterogeneity within tumors, characterization of metabolic programs (metabolic flux distributions) at the single-cell level is required

• we propose to integrate cancer data from: 1) single-cell transcriptomics and 2) bulk metabolomics, into a multi-scale stoichiometric model

• An advantage of FBA is that, as opposed to intracellular fluxes, extracellular fluxes can also be approximated from measurements of the concentration of metabolites in the spent cell culture medium at different time points.

• we cannot assume that each cell within an in vivo cancer population proliferates at the same rate, nor that it proliferates at all

• we do not impose that the population dynamics is at steady-state

• scFBA aims at portraying a snapshot of the single-cell (steady-state) metabolic phenotypes within an (evolving) cell population at a given moment, and at identifying metabolic subpopulations, without a priori knowledge, by relying on unsupervised integration of scRNA-seq data.

• At the implementation level, we use continuous data, rather than discrete levels, to overcome the problem of selecting arbitrary cutoff thresholds

• we use instead the “pipe capacity” philosophy embraced by other methods, such as the E-Flux method [36, 37], of setting the flux boundaries as a function of the expression state

• we prefer to normalize boundaries in relation to the condition/cell/tissue in which a given reaction is mostly expressed, as done in a more recent version of the E-flux method

• we distribute the total (bulk) possible flux of each reaction proportionally to the activity score of that reaction in each cell

• The risk of the presence of false negatives in RNA-seq, and in particular scRNA-seq, is an established problem. Although a totally safe solution does not exist, scFBA allows to employ the information on bulk expression profile, when available, to manage the risk, by envisioning the following scenarios.

• If a gene has a zero read count in the bulk, as well as in each single-cell, we cannot totally exclude the possibility of a false-negative in the bulk, but we are confident in excluding a false-negative due to low concentrations of scRNA-seq, thus we can assume that such gene is off in all cells

• If a gene has non-zero read count in the bulk, but a zero read count in each single-cell, there is a sharp inconsistency between bulk and scRNA-seq that indicates that we cannot trust scRNA-seq for this gene. In this situation, we prefer to lose information on single-cell heterogeneity and rely on the bulk value: we replace the read count for that gene in each cell with the bulk read count.