Closed Yijun-Tian closed 1 year ago
Hello @Yijun-Tian,
If you're concerned that there are edge effects in your data, we recommend using modkit extract
to hone in on the details of that effect. However, in typical runs, however 10bp is likely quite safe. This will remove any false canonical calls due to adapter ligation and most or all edge effects in the signal. For very long reads you may want to increase the --edge-filter
up to as much as 30bp (which was used in megalodon previously).
Thanks for the clarification. Does this rule also apply to the adaptive sampling reads? As the adaptive sampling reads undergoes frequent rejecting or reversing processes, should I use longer --edge-filter in this case?
@Yijun-Tian For the accepted reads during Adaptive Sampling, I would not expect that you'd need to treat them any differently that typical reads. For the rejected reads I'd recommend looking at the output of modkit extract
to decide how best to handle them (if you're going to use them at all).
Hi modkit maker, For the modkit summary module, do you recommend a rough number for the --edge-filter option when processing common genomic sequencing reads? Is there any literature discussing why or when these edge-trimming is needed? Thank you,