Fig. 1: Extended library development, samples intended analysis and expected and optional measures from the library application. | Nature Communications

Fig. 1: Extended library development, samples intended analysis and expected and optional measures from the library application.

From: Enhanced cell deconvolution of peripheral blood using DNA methylation for high-resolution immune profiling

Fig. 1

a Twelve cell types were acquired commercially, their DNA was isolated, and DNA methylation was measured using the Illumina HumanMethylationEPIC (EPIC) microarray. Using artificial mixtures as ground truth, two libraries were identified using an iterative process named IDOL (IDentifying Optimal Libraries, Koestler, et al. 2016). The two extended (ext) libraries were designed for microarray data derived from the EPIC array (EPIC IDOL-ext) or legacy data derived from the previous Illumina-HumanMethylation450k array (450k IDOL-ext). b Samples with variable amounts of leukocytes are arrayed using any of the two validated microarray technologies. Using the appropriate library for the microarray, a cell mixture deconvolution is performed using the constrained projection/quadratic programming (CP/QP, Houseman, et al. 2012). *Optionally, leukocyte counts can be collected for downstream analyses. c The primary results of the deconvolution are the 12 cell types of the library. These results could be aggregated at different levels for different hypotheses. Two sets of derived results are possible: (1) if total leukocyte counts are available cell-type-specific counts may be inferred, or (2) cell ratios and proportions are used to evaluate immune-cell shifting between the different cell-type subpopulations (only a few examples illustrated here).

Back to article page