Abstract
CpG methylation is an important epigenetic feature that plays a key role in gene expression, and changes in methylation patterns have been implicated in the development and progression of several diseases. Methylation profiling is therefore a highly important technique that enables the identification of methylated sites and has widespread uses in biomedical research. The initial development of bisulfite conversion as a technique to distinguish between unmethylated and methylated cytosines was significant in this field and provided a basis for the continued improvement of library preparation techniques, with bisulfite conversion remaining to be the most used technique for producing methylome libraries. However, the conversion process results in DNA damage and biases which have significant implications for the quality of libraries. These biases are more pronounced in lower DNA inputs, due to an increase in the number of PCR cycles needed. This has led to the exploration of bisulfite-free methods of obtaining methylation libraries, such as EM-seq (NEBNext Enzymatic Methyl-seq), intending to overcome issues that arise due to bisulfite-induced DNA damage. The performance of EM-seq libraries has been previously shown to be promising and indicates potential for improved strength of methylation libraries in future research, however there is little literature which evaluates the performance of this technique at low inputs. Using an A549 cell line, libraries were prepared using both a bisulfite based technique; PICO methyl-seq, and EM-seq. Libraries prepared with PICO methyl-seq kit showed variability both within and between preparations, while EM-seq showed more consistent results across preparations. Both conversion techniques were able to produce sufficient bioanalyzer traces from inputs of 100ng to 0.1ng, and EM-seq libraries showed a larger fragment size than PICO methyl-seq libraries. Yield significantly lower in EM-seq prepared libraries than PICO Methyl-seq prepared libraries, and they required more PCR cycles in order to produce a quantifiable yield at lower inputs of 1 and 0.1ng. This indicates that PICO methyl-seq may be more suitable for these low input sizes, however further sequencing and analysis of the libraries across metrics such as GC bias and coverage depth would be vital in understanding the impact that the number of PCR cycles has on the quality of the methylation data. Additionally, the results indicate a future avenue for optimisation of the EM-seq protocol for low inputs.