DeconPeaker, a Deconvolution Model to Identify Cell Types Based on Chromatin Accessibility in ATAC-Seq Data of Mixture Samples

Over the years, our understanding towards the cellular and molecular processes has grown exponentially. At the same time, the issues related to cell microenvironment and cellular heterogeneity open a new debate regarding the cell identity. After all, it is the cell composition (chromatin and nuclear architecture) which poses a strong risk for dynamic changes in the diseased condition. Since, chromatin accessibility patterns play a major role in the human diseases, hence, a deconvolution tool based on open chromatin data is expected to provide better performance in identifying cell composition. Herein, we designed the deconvolution tool “DeconPeaker”, which can precisely define the uniqueness among subpopulations of cells by using open chromatin datasets. Using this tool, we simultaneously evaluated chromatin accessibility datasets and gene expression datasets to estimate the cell types and their respective proportions in the mixture of samples. In comparison to other known deconvolution methods, we observed the lowest average root-mean-square error (RMSE = 0.042) and the highest average correlation coefficient (r = 0.919) between the prediction and the “true” proportion. As a proof of concept, we also tested chromatin accessibility data from acute myeloid leukemia (AML) and successfully obtained the unique cell types related to the progression for AML. Also, we could show that chromatin accessibility represents more essential characteristics than gene expression in identifying the cell types. Taken together, DeconPeaker as a powerful tool has a potential to combine different datasets (chromatin accessibility and gene expression) and define the different cell types from mixtures, and an R-library is available on https://github.com/lihuamei/LinDeconSeq.