I'm hoping that somebody here can explain this unusual result that I'm getting with some pyrosequencing data. I have bisulphite converted a few samples a couple of times over the years for different projects. Recently I was collecting some of this disparate data together and I noticed that when the same pyrosequencing assay was used on these samples, those that were converted later had decreased methylation at the CpG sites of interest compared to the initial bisulphite converted samples. The original conversion was in 2010 and the repeat conversions were in 2014/2015. The assays are the same and it's difficult to blame this on PCR amplifying one clone over another since I've done technical replicates on each aliquot. Even the 2010 converted sample has the same result in 2015 that it did in 2010 but this is different from the 2015 converted sample.
We are using the same kit (Qiagen) for bisulphite conversion and haven't made a major change to our procedure (we follow the Qiagen protocol without modification). Has anybody noticed this phenomenon before? Is it possible that stored DNA would experience methylation changes? It seems more likely that the issue is technical but I'm having trouble figuring out exactly what that technical issue is.