search

bioinfomaticsCSU / deepsignal

Detecting methylation using signal-level features from Nanopore sequencing reads
GNU General Public License v3.0
108 stars 21 forks source link

Do I need Whole Genome Amplified dataset? #81

Open shailabhr opened 2 years ago

shailabhr commented 2 years ago

Hi, I have three bacterial whole genome sequencing samples in which the first is control and the other two are the same organism grown in different growth environments. I have sequenced all three samples together in the same library with different barcodes. I do not have the WGA dataset. My aim is to detect the changes in methylation in my two samples as compared to the control. I did my experiments before realizing that many tools need a WGA dataset to analyze methylation data. Does Deepsignal need it too? Your paper says you have used WGA data to train your model. I just wanted to confirm if I will need WGA dataset if I plan to analyze my data with Deepsignal. Thank you for your help. Sincerely, Shailabh

PengNi commented 2 years ago

Hi @shailabhr , thank you very much for your interest of deepsignal. It is indeed that deepsginal does't need WGA dataset.However, I don't think deepsignal will perform well in modern ONT R9.4+ bacterial datasets, as the deepsignal CCWGG and 6mA models were trained using old ONT R9 datasets.

If you want analyze CCWGG or 6mA in bacterial genomes, I suggest you check the ONT Remora/Megalodon tools.

Best, Peng

shailabhr commented 1 year ago

Thank you @PengNi . Would you recommend using DeepSignal2 instead? I am trying to use Remora in parallel.

PengNi commented 1 year ago

@shailabhr , I think maybe remora/mcaller is more suitable.