search

sartorlab / methylSig

R package for DNA methylation analysis
17 stars 5 forks source link

DMR results and methylation level #53

Open chantisakee opened 1 year ago

chantisakee commented 1 year ago

Hi,

I got DMR results from the MethylSig model. then i extracted methylation levels of each sample based on different methylated regions. I just wonder why there are some significant regions with zero methylation.

ex. this is DMR from methylsig model: bin no. 778 and 817 778,chr1,778001,779000,1000,,0.741841384941123,2.62058969173205,-1.87874830679093,Normal,0.0218720395180972,0.854065540068584,61.4371023713012,6.4479341674014,16 817,chr1,818001,819000,1000,,87.2264615043364,82.6203527665182,4.60610873781825,Cancer,0.0375882786613783,0.854065540068584,232.033232163809,5.14017691340996,16

i extracted this bin no. from bsseq::getCoverage(bs_input, type = 'M') ; 778,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 817,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0

Thank you, Chantisa

rcavalcante commented 1 year ago

Hi,

That is sort of bizarre. Is it possible to share your BSseq object with me so I can look into it? Also, could you please let me know your sessionInfo()?

Thanks, Raymond

chantisakee commented 1 year ago

Hi, thank you for your response! :) This is what I’ve done and what I’ve found so for. I created a binning windows windowed_bs = tile_by_windows(bs = bs_input, win_size = 1000)

  1. Used this bining windows as the input in the model (in this case - methylsig model)

diff_gr_methylSig = diff_methylsig( bs = windowed_bs, group_column = 'Type', comparison_groups = c('case' = 'Group2', 'control' = 'Group1'), disp_groups = c('case' = TRUE, 'control' = TRUE), local_window_size = 0, t_approx = TRUE, n_cores = 4)

  1. Also, used this bining windows as the input in order to get “methylation profile”

bsseq::getCoverage(windowed_bs, type = 'M')

And I used row no. From 2# output matching with row no. From 3# output and the result is wrong (like there are significant region in 2# but methylation profiles are all zeroes in 3#

But I found out that if I used “binominal model”, row no. Of 2# output will be equal to 3#. (Since number of rows in 2# are equal to #3). Whereas, methylsig model and dss model are not. So, is it possible to get methylation profile with DMR output from each model instead of getting from bsseq::getCoverage(windowed_bs, type = 'M'). That might be helpful.

On 7 Mar BE 2566, at 22:29, Raymond Cavalcante @.***> wrote:

 Hi,

That is sort of bizarre. Is it possible to share your BSseq object with me so I can look into it? Also, could you please let me know your sessionInfo()?

Thanks, Raymond

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.