Benchmarking sample pooling for epigenomics of natural populations

Abstract

Interest in the role of DNA methylation (DNAm) has grown in ecological and evolutionary research of natural populations. While researchers are typically interested in comparing population-level variation, individual sequencing is the current standard. Natural populations have low effect sizes and thus need large sample sizes to detect differences. The cost of sequencing the necessary samples can be prohibitive in DNAm work. Pooling DNA before library preparation is a powerful tool to reduce costs but no recommendations exist for DNAm pooling in ecology-epigenetics research. We test if pooled and individual libraries provide similar global and region-specific DNA methylation signals in a natural system of response to pollution. We generated whole-epigenome data for two freshwater invasive molluscs (Corbicula flumina and Dreissena polymorpha) collected from a polluted and unpolluted locality, Lake Maggiore, Italy. Our results support that pooling effectively captures the same genome-wide and global treatment-level signals as individual libraries but we note that pooled libraries yielded orders of magnitude more input data and differentially-methylated regions (DMRs) detected compared with individual libraries. We estimated greatly lower power for regions from individual libraries compared with pooled libraries. The post-hoc process of computationally pooling data from individual libraries produced results comparable to pooled libraries in volumes but had discrepancies between DMRs. We discuss the possible causes for the discrepancies and put our results in the context of the benefits and drawbacks of sample pooling for epigenomics of natural populations.