BayesCCE: a Bayesian framework for estimating cell-type composition from DNA methylation without the need for methylation reference

Abstract

We introduce a Bayesian semi-supervised method for estimating cell counts from DNA methylation by leveraging an easily obtainable prior knowledge on the cell-type composition distribution of the studied tissue. We show mathematically and empirically that alternative methods which attempt to infer cell counts without methylation reference only capture linear combinations of cell counts rather than provide one component per cell type. Our approach allows the construction of components such that each component corresponds to a single cell type, and provides a new opportunity to investigate cell compositions in genomic studies of tissues for which it was not possible before.Electronic supplementary materialThe online version of this article (10.1186/s13059-018-1513-2) contains supplementary material, which is available to authorized users.

Cell-type-specific resolution epigenetics without the need for cell sorting or single-cell biology

“…Importantly, TCA requires knowledge of the cell-type proportions of the individuals in the data. These can be computationally estimated using either a reference-based supervised approach 27 or a reference-free semi-supervised approach 28 ; current reference-free unsupervised methods, however, are unable to provide reasonable estimates of cell-type proportions but rather only linear combinations of them 28 . Notably, in cases where only noisy estimates of the cell-type proportions are available (i.e.…”

“…A potential limitation of TCA is the need for rarely available cell-type proportions as an input. We alleviate this issue by allowing TCA to get estimates of the cell-type proportions using standard methods 27,28 and then re-estimating them following the TCA model. As we showed, this allows TCA to provide good results even when just noisy estimates of the cell-type proportions are available.…”

“…As we showed, this allows TCA to provide good results even when just noisy estimates of the cell-type proportions are available. In practice, obtaining such estimates can be done using either a reference-based approach 27 or a semi-supervised approach 28 , in case a methylation reference is not available for the studied tissue.…”

Distinguishing biological from technical sources of variation by leveraging multiple methylation datasets

“…In an attempt to overcome the above limitations, many reference-free methods [21][22][23][24][25][26][27], have been proposed. These methods attempt to find a linear transformation of the variability of interest, and use this transformed version of the signal as a surrogate or covariate to control for their effects in EWAS or other analyses.…”

“…These methods attempt to find a linear transformation of the variability of interest, and use this transformed version of the signal as a surrogate or covariate to control for their effects in EWAS or other analyses. Though these methods can correct for cell-type composition in EWAS [25,28,29] and may also capture other sources of variability, they are limited by the fact that it is impossible to know whether their components reflect biological or technical signal (Figure 1). While technical signal is not of interest and should be accounted for in the analysis, the biological signal can provide useful insights about underlying biological phenomena, for instance by being used to model the interaction with the methylation signal.…”