Confounding due to cellular heterogeneity represents one of the major analytical challenges in Epigenome-Wide Association Studies (EWAS) conducted on biospecimens comprised of mixed cell populations (e.g., whole blood). Statistical methods leveraging the tissue-specificity of DNA methylation for deconvolving the cellular mixture of heterogenous biospecimens (so-called deconvolution methods) offer a promising solution; consequently, such methods have become widely accepted as essential for statistically rigorous EWAS. There have been significant advancements in DNA methylation-based deconvolution methods over the past decade and even though the original paper describing deconvolution in the context of DNA methylation data (the so-called Houseman method) was published over a decade ago, deconvolution remains an active area of statistical and computational research. In this talk, I will describe the statistical underpinnings of DNA methylation-based deconvolution along with discussing some of the significant advancements that have taken place since the publication of the Houseman method in 2012. Recent work on the development of novel compositional data analysis methods and their application to DNA methylation-based deconvolution estimates will also be described.