Entering edit mode
7.3 years ago
t2g4free
•
0
Hi all, I have some question about WGBS.
how to get the percentage of cytosines of bisulphite-treated reads?
If i have only 2 sample and no replicate, which software is better to get DMR?
what on earth the coverage and depth is, and how to calculate them?
for 2, I read some paper, they use different statistical methods, Fish exact test, t-test, binominal, beta-binomial... hash, they all confuse me a lot. Different paper say different conclusion, but which one suite my condition?
Have a look at the Bismark alignment tool and the R/Bioconductor package methylKit. Bismark alignes the data to an in silico converted genome and methylKit provides an analysis framework. Use something published, never build statistics yourself unless you are an expert in it.
Still, especially question 3 indicates that you have little or no NGS background. Better start with some basics before jumping into genome-wide analysis. You need to be familiar with alignments, data formats and file processing (indexing, sorting, deduplicating). In WG(B)S also the sizes of the files are an obstacle, so some basics in using Unix pipes for sure helps, unless you have extensive disk space.
Thx 4 your reply. the 3ed question is different in different article, so I was confused.
Yeah, some good advice. for the 1st question, I prefer to know the percent of C that are coverage by the bisulphite-treated reads in genome wide.
As I said, align them with a bisulfite-aware aligner, such as bismark, then extract the information with methylKit. Both tools have good a documentation, so it should not be a problem.
Hi, for the 2nd, you can try the absolute difference larger than 0.3. If you have more samples without replicates, you can try the Entropy-based methods such as QDMR (http://fame.edbc.org/qdmr/) or SMART (http://fame.edbc.org/smart/), both of them can process such kind of situation.