search

al-mcintyre / mCaller

A python program to call methylation (m6A in DNA) from nanopore signal data
MIT License
45 stars 16 forks source link

model training with my own nanopore data #40

Open dashengzhao opened 6 months ago

dashengzhao commented 6 months ago

Hello, Dr McIntyre, Althrough some research have used PacBio data to detect 6mA in the plant which l' m interested, using nanopore data to detect 6mA in this plant species has not been available so far. So, i want to first detect genome-wide " true" 6mA sites in this plant species using my own experimental nanopore data with extremely strict parameters (such as : make_bed.py -d 30 -t 0.9 ), and then using the outputs to train the original r94_model_NN_6_m6A.pkl to get a new model, and finally return to use this new model to detect genome-wide 6mA site in this plant species using my experimental nanopore data with suggested parameters (such as : make_bed.py -d 15 -t 0.5 ). Is this idea feasible ? Could you give some suggestions ?

Thank you.

al-mcintyre commented 6 months ago

Can you use the sites predicted by PacBio to train the model? Does methylation tend to be binary for single sites (ie. ~100% or 0% of reads)?