Normalization of the WGBS smooth data

[priyatamapandey]

Hi,
How can I normalize my smooth data?

Thank you for the help,
Priya

[priyatamapandey]

Hi,
I am finding methylation signal low in some of the my samples and suspecting batch effect and want to normalize the smooth beta values (returned after getMeth() function). Want to feed this matrix into the object again but as bsseq object does not keep smoothed matrix directly. It looks to me it keep coef matrix which transform in smooth data. So I am unable to find the way where I can replace my smooth methylated profile/matrix with normalize smooth profile as I want to feed this object file in the DMRcate tool to identify DMRs.
Looking forward for suggestions.
Thank you,
Priya

[PeteHaitch]

I've never done any normalization of the data following smoothing.
@kasperdanielhansen have you?

As you note, the BSseq object does not contain the smoothed values, only the parameters, but the smoothed values can be obtained with getMeth().
I don't know how DMRcate works, but perhaps there's a way to pass in (normalized) values of getMeth() as input rather than a BSseq object to the DMRcate functions you want to use?
That's a question for the developers of that package.

[kasperdanielhansen]

It would be helpful to get some more details of what you want. I have never normalized the data following smoothing, but I have included covariates in a linear model, to account for systematic shifts. For example, this is one way to address a possible change in genome-wide average methylation which I have (one time) seen occur as a batch effect. Best, Kasper

…

On Tue, Sep 13, 2022 at 6:28 PM Peter Hickey ***@***.***> wrote: I've never done any normalization of the data following smoothing. @kasperdanielhansen <https://github.com/kasperdanielhansen> have you? As you note, the *BSseq* object does not contain the smoothed values, only the parameters, but the smoothed values can be obtained with getMeth(). I don't know how *DMRcate* works, but perhaps there's a way to pass in (normalized) values of getMeth() as input rather than a *BSseq* object? That's a question for the developers of that package. — Reply to this email directly, view it on GitHub <#114 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABF2DHZR54OXRKQWFNWTZWLV6D5YLANCNFSM57BZITSA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Best, Kasper

[priyatamapandey]

Thank you for considering my question. I have visualized a DMR region of the smooth data and suspecting the batch effect. I want to identify drms using the smooth quantile normalized data with DMRcate as I have many covariates.
I checked DMRcate uses bsseq object not any data frame/matrix.

Here is the screenshot I am sending for understanding..

Thank you for any help and suggestions.
Priya

[PeteHaitch]

Sorry, I don't understand what you're trying to show with that screenshot.