Optimal Deconvolution of Transcriptional Profiling Data Using Quadratic Programming with Application to Complex Clinical Blood Samples

Abstract

Large-scale molecular profiling technologies have assisted the identification of disease biomarkers and facilitated the basic understanding of cellular processes. However, samples collected from human subjects in clinical trials possess a level of complexity, arising from multiple cell types, that can obfuscate the analysis of data derived from them. Failure to identify, quantify, and incorporate sources of heterogeneity into an analysis can have widespread and detrimental effects on subsequent statistical studies.We describe an approach that builds upon a linear latent variable model, in which expression levels from mixed cell populations are modeled as the weighted average of expression from different cell types. We solve these equations using quadratic programming, which efficiently identifies the globally optimal solution while preserving non-negativity of the fraction of the cells. We applied our method to various existing platforms to estimate proportions of different pure cell or tissue types and gene expression profilings of distinct phenotypes, with a focus on complex samples collected in clinical trials.We tested our methods on several well controlled benchmark data sets with known mixing fractions of pure cell or tissue types and mRNA expression profiling data from samples collected in a clinical trial. Accurate agreement between predicted and actual mixing fractions was observed. In addition, our method was able to predict mixing fractions for more than ten species of circulating cells and to provide accurate estimates for relatively rare cell types (<10% total population). Furthermore, accurate changes in leukocyte trafficking associated with Fingolomid (FTY720) treatment were identified that were consistent with previous results generated by both cell counts and flow cytometry. These data suggest that our method can solve one of the open questions regarding the analysis of complex transcriptional data: namely, how to identify the optimal mixing fractions in a given experiment.

Early Pregnancy Maternal Blood DNA Methylation in Repeat Pregnancies and Change in Gestational Diabetes Mellitus Status—A Pilot Study

“…Advanced procedures being developed in other areas of research (eg, transcriptomics) that help determine cell content of samples will mitigate this concern in future studies. 29,30 Finally, the generalizability of findings needs to be enhanced by conducting similar experiments in more diverse study populations.…”

A Critical Survey of Deconvolution Methods for Separating Cell Types in Complex Tissues

“…An extension of this approach is proposed by Qiao et al [21], which uses non-negative least squares (NNLS) to explicitly enforce non-negativity as part of the optimization. Gong et al [22] present a quadratic programming (QP) framework to explicitly encode both constraints in the optimization problem formulation. They also propose an extension to this method, called DeconRNASeq, which applies the same QP framework to RNASeq datasets.…”

“…• BreatBlood [22] (GEO ID: GSE29830): Breast and blood from human specimens are mixed in three different proportions and each of the mixtures is measured three times, with a total of nine samples.…”

“…For example, Gong et al [22] suggest an effective range of [0.5, 5000], whereas Ahn et al [43] observe an optimal range of [2 4 − 2 14 ]. To facilitate the choice of expression bounds, we seek a systematic way to identify an optimal range for different datasets.…”

Comprehensive Evaluation of RNA-seq Quantification Methods for Linearity

“…In the field of biomedical research, deconvolution is widely applied to retrieve cell-type or tissue specific gene expression profiles from heterogeneous tissue samples. Most deconvolution algorithms in the literature assume a linear model [10][11][12][13][14][15][16][17], in which the expression signal of the mixture is a weighted sum of the expression for its constitutive cell types. Previous analysis has shown the necessity of using anti-log expression microarray data to avoid unwanted bias introduced by non-linear transformation [18].…”