Hi there,

We have been trying to use the EM-seq from NEB to construct libraries. When using lambda DNA without 5mC as a model DNA, we found the conversion rate is usually less than 99%. The sequencing data showed that the unconverted site distributions are not random. For most of the reads, the conversion rates are very high, however, for ~5-10% reads, more than half of the C sites in these reads kept unconverted. When we filtered these reads with a criteria that >50% of C unconverted were filtered, the conversion rate can reach >99%, but the filtration step is not a routine step in DNA 5mC detection although it is used to remove clustered reads caused by the secondary structures in m5C detection in RNA . This suggested that the denature step before APOBEC treatment is not adequate to convert all the dsDNA to ssDNA. We have tried both methods for denaturation in the protocol (using 0.02 M NaOH or form amide at 85 C for 10 min), but the same issue cannot be solved. We also tested human cfDNA, and the methylation level detected in CHG and CGG motifs are much higher than BS-seq method. I am wondering whether you have met this situations before. Any suggestions to solve the problem are welcome.

Besides, anyone have tested EM-seq with lower than 10 ng input, such as 1 ng, or 0.1 ng? What is the conversion rate for such low input samples?

In our sequencing data, we also found 0.3 to 3% of reads can be mapped to E. Coli, the E. Coli contamination rate is EM-seq kit batch dependent, not sure whether the contamination came for the Test/APOBEC enzyme, or from the environment in the course of the experiments.

In addition, I also have concerns that TET may bind to the dsDNA substrate tightly and it cannot be removed in the purification step after TET oxidation, which may protect the DNA from deamination in APOBEC treatment. Is this possible? Thanks.