Abstract
BackgroundCell-type heterogeneity of tumors is a key factor in tumor progression and response to chemotherapy. Tumor cell-type heterogeneity, defined as the proportion of the various cell-types in a tumor, can be inferred from DNA methylation of surgical specimens. However, confounding factors known to associate with methylation values, such as age and sex, complicate accurate inference of cell-type proportions. While reference-free algorithms have been developed to infer cell-type proportions from DNA methylation, a comparative evaluation of the performance of these methods is still lacking.ResultsHere we use simulations to evaluate several computational pipelines based on the software packages MeDeCom, EDec, and RefFreeEWAS. We identify that accounting for confounders, feature selection, and the choice of the number of estimated cell types are critical steps for inferring cell-type proportions. We find that removal of methylation probes which are correlated with confounder variables reduces the error of inference by 30–35%, and that selection of cell-type informative probes has similar effect. We show that Cattell’s rule based on the scree plot is a powerful tool to determine the number of cell-types. Once the pre-processing steps are achieved, the three deconvolution methods provide comparable results. We observe that all the algorithms’ performance improves when inter-sample variation of cell-type proportions is large or when the number of available samples is large. We find that under specific circumstances the methods are sensitive to the initialization method, suggesting that averaging different solutions or optimizing initialization is an avenue for future research.ConclusionBased on the lessons learned, to facilitate pipeline validation and catalyze further pipeline improvement by the community, we develop a benchmark pipeline for inference of cell-type proportions and implement it in the R package medepir.
Highlights
Cell-type heterogeneity of tumors is a key factor in tumor progression and response to chemotherapy
All these software packages perform various types of non-negative matrix factorization to infer cell type proportions and cell type-specific methylation profiles by solving D = TA, or rather by minimizing, under various constraints, the error term: ∥D − TA∥2
We simulate D with 5 cell types (K = 5): 2 cancer-like cells, healthy epithelial cells, immune cells (T lymphocytes), and stromal cells. These simulations mainly depend on a parameter α0, which controls the diversity of the generated samples: When α0 is small (~ 1), the simulated proportions of the K cell-types are diverse among samples and as α0 increases, the variability decreases to the point at which proportions are the same for all samples
Summary
Cell-type heterogeneity of tumors is a key factor in tumor progression and response to chemotherapy. While reference-free algorithms have been developed to infer cell-type proportions from DNA methylation, a comparative evaluation of the performance of these methods is still lacking. The “reference-free” algorithms are labelled as such because they do not require a priori information about DNA methylation profiles of cell types found within tumors: they directly infer them from DNA methylation samples using computational methods. In the absence of reference information, confounding factors affecting methylation values, such as age and sex, can potentially influence the inference of cell-type proportions. There is an urgent need to comprehensively characterize the current analysis pipelines, identify key features influencing their performance and provide benchmarks and recommendations to guide the application and further development of pipelines that quantify cell-type heterogeneity from reference-free DNA methylation samples [6]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
