Benchmark of cellular deconvolution methods using a multi-assay reference dataset from postmortem human prefrontal cortex

Background

Cellular deconvolution of bulk RNA-sequencing (RNA-seq) data using single cell or nuclei RNA-seq (sc/snRNA-seq) reference data is an important strategy for estimating cell type composition in heterogeneous tissues, such as human brain. Computational methods for deconvolution have been developed and benchmarked against simulated data, pseudobulked sc/snRNA-seq data, or immunohistochemistry reference data. A major limitation in developing improved deconvolution algorithms has been the lack of integrated datasets with orthogonal measurements of gene expression and estimates of cell type proportions on the same tissue sample. Deconvolution algorithm performance has not yet been evaluated across different RNA extraction methods (cytosolic, nuclear, or whole cell RNA), different library preparation types (mRNA enrichment vs. ribosomal RNA depletion), or with matched single cell reference datasets.

Results

A rich multi-assay dataset was generated in postmortem human dorsolateral prefrontal cortex (DLPFC) from 22 tissue blocks. Assays included spatially-resolved transcriptomics, snRNA-seq, bulk RNA-seq (across six library/extraction RNA-seq combinations), and RNAScope/Immunofluorescence (RNAScope/IF) for six broad cell types. The Mean Ratio method, implemented in the DeconvoBuddies R package, was developed for selecting cell type marker genes. Six computational deconvolution algorithms were evaluated in DLPFC and predicted cell type proportions were compared to orthogonal RNAScope/IF measurements.

Conclusions

Bisque and hspe were the most accurate methods, were robust to differences in RNA library types and extractions. This multi-assay dataset showed that cell size differences, marker genes differentially quantified across RNA libraries, and cell composition variability in reference snRNA-seq impact the accuracy of current deconvolution methods.

Performance of tumour microenvironment deconvolution methods in breast cancer using single-cell simulated bulk mixtures

Cells within the tumour microenvironment (TME) can impact tumour development and influence treatment response. Computational approaches have been developed to deconvolve the TME from bulk RNA-seq. Using scRNA-seq profiling from breast tumours we simulate thousands of bulk mixtures, representing tumour purities and cell lineages, to compare the performance of nine TME deconvolution methods (BayesPrism, Scaden, CIBERSORTx, MuSiC, DWLS, hspe, CPM, Bisque, and EPIC). Some methods are more robust in deconvolving mixtures with high tumour purity levels. Most methods tend to mis-predict normal epithelial for cancer epithelial as tumour purity increases, a finding that is validated in two independent datasets. The breast cancer molecular subtype influences this mis-prediction. BayesPrism and DWLS have the lowest combined numbers of false positives and false negatives, and have the best performance when deconvolving granular immune lineages. Our findings highlight the need for more single-cell characterisation of rarer cell types, and suggest that tumour cell compositions should be considered when deconvolving the TME.

Toward Precision Radiotherapy: A Nonlinear Optimization Framework and an Accelerated Machine Learning Algorithm for the Deconvolution of Tumor-Infiltrating Immune Cells

Tumor-infiltrating immune cells (TIICs) form a critical part of the ecosystem surrounding a cancerous tumor. Recent advances in radiobiology have shown that, in addition to damaging cancerous cells, radiotherapy drives the upregulation of immunosuppressive and immunostimulatory TIICs, which in turn impacts treatment response. Quantifying TIICs in tumor samples could form an important predictive biomarker guiding patient stratification and the design of radiotherapy regimens and combined immune-radiation treatments. As a result of several limitations associated with experimental methods for quantifying TIICs and the availability of extensive gene sequencing data, deconvolution-based computational methods have appeared as a suitable alternative for quantifying TIICs. Accordingly, we introduce and discuss a nonlinear regression approach (remarkably different from the traditional linear modeling approach of current deconvolution-based methods) and a machine learning algorithm for approximating the solution of the resulting constrained optimization problem. This way, the deconvolution problem is treated naturally, given that the gene expression levels of pure and heterogenous samples do not have a strictly linear relationship. When applied across transcriptomics datasets, our approach, which also allows the coupling of different loss functions, yields results that closely match ground-truth values from experimental methods and exhibits superior performance over popular deconvolution-based methods.

Robust and accurate estimation of cellular fraction from tissue omics data via ensemble deconvolution

MOTIVATION

Tissue-level omics data such as transcriptomics and epigenomics are an average across diverse cell types. To extract cell-type-specific (CTS) signals, dozens of cellular deconvolution methods have been proposed to infer cell-type fractions from tissue-level data. However, these methods produce vastly different results under various real data settings. Simulation-based benchmarking studies showed no universally best deconvolution approaches. There have been attempts of ensemble methods, but they only aggregate multiple single-cell references or reference-free deconvolution methods.

RESULTS

To achieve a robust estimation of cellular fractions, we proposed EnsDeconv (Ensemble Deconvolution), which adopts CTS robust regression to synthesize the results from 11 single deconvolution methods, 10 reference datasets, 5 marker gene selection procedures, 5 data normalizations and 2 transformations. Unlike most benchmarking studies based on simulations, we compiled four large real datasets of 4937 tissue samples in total with measured cellular fractions and bulk gene expression from different tissues. Comprehensive evaluations demonstrated that EnsDeconv yields more stable, robust and accurate fractions than existing methods. We illustrated that EnsDeconv estimated cellular fractions enable various CTS downstream analyses such as differential fractions associated with clinical variables. We further extended EnsDeconv to analyze bulk DNA methylation data.

AVAILABILITY AND IMPLEMENTATION

EnsDeconv is freely available as an R-package from https://github.com/randel/EnsDeconv. The RNA microarray data from the TRAUMA study are available and can be accessed in GEO (GSE36809). The demographic and clinical phenotypes can be shared on reasonable request to the corresponding authors. The RNA-seq data from the EVAPR study cannot be shared publicly due to the privacy of individuals that participated in the clinical research in compliance with the IRB approval at the University of Pittsburgh. The RNA microarray data from the FHS study are available from dbGaP (phs000007.v32.p13). The RNA-seq data from ROS study is downloaded from AD Knowledge Portal.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

How many markers are needed to robustly determine a cell's type?

Our understanding of cell types has advanced considerably with the publication of single-cell atlases. Marker genes play an essential role for experimental validation and computational analyses such as physiological characterization, annotation, and deconvolution. However, a framework for quantifying marker replicability and selecting replicable markers is currently lacking. Here, using high-quality data from the Brain Initiative Cell Census Network (BICCN), we systematically investigate marker replicability for 85 neuronal cell types. We show that, due to dataset-specific noise, we need to combine 5 datasets to obtain robust differentially expressed (DE) genes, particularly for rare populations and lowly expressed genes. We estimate that 10 to 200 meta-analytic markers provide optimal downstream performance and make available replicable marker lists for the 85 BICCN cell types. Replicable marker lists condense interpretable and generalizable information about cell types, opening avenues for downstream applications, including cell type annotation, selection of gene panels, and bulk data deconvolution.