Asymmetric methylation

[Shivani Singh]

Can we also analyse asymmetric pattern of cystosine methylation (in case of plants) apart from CpG using methylCRF algorithm ??

[michael]

Hi Shivani,

The particular model we trained doesn't predict asymmetric methylation.  However, the statistical method can be used to train a model that predicts asymmetic methylation. 

The method is quit general in this way -you can add or subtract any number of features.  However, there is more work involved to train new models over simply using a trained one.  It requires some bioinformatics skills as training involves multiple steps that aren't coherently wrapped up in a turn-key type app.  Also, any new model with a different combination of features would have to be somehow validated or characterized to determine its accuracy.   

For asymmetric methylation, though, you could just try running 5' and 3' strands seperately -although, again, this hasn't been validated.

best,
michael

[Shivani Singh]

Thanks for your response :)
Actually I also want to know, as in asymmetric methylation we are not sure about combination of nucleotide either it will be CHG or CHH ( H = A,T,G). So, how can we modulate the algorithm to meet our requirement ?