We are doing a genome-wide analysis of methylation via bisulfite sequencing for an insect species. Previous experimental techniques have suggested the presence of methylation in this organism, including an AFLP analysis using methylation-sensitive restriction enzymes and a differential expression analysis using a methylation inhibitor.
Our bisulfite sequencing data, however, is giving us essentially no evidence of methylation. If this finding is true, it will be a pretty big deal. But before we go down that path, we want to make sure we've counted out all sources of technical error.
First, we wondered if the bisulfite treatment failed, but quickly determined that this would not explain the result. If the C --> T conversion had failed at an extremely high rate, you would expect to see methylation all over the place. That's the opposite of what we're seeing, which is essentially no methylation.
Second, we wondered if there was some kind of "overexposure" issue in the bisulfite treatment that caused even methylated Cs to convert to Ts. We found a paper describing this phenomenon[1], but reports we can find suggest this happens at a low (<10%) rate, which cannot explain our data.
Third, we wondered if perhaps there was an issue with the reads mapping to the genome, we got at least 70% mapping efficiency for each BS-seq sample, so that can't be it either.
Are there any other sources of technical error we should consider and check for before believing this result?
