Filter fail reads with Megalodon vs Remora

[Coracollar]

Hi,
I've realized megalodon does not filter out fail reads after base calling even though the config file does have a pass filter. So I was wondering:

1. Is there a specification on the main megalodon command that filters out fail reads?

2. Alternatively, I've found a way of taking them out of the per_read_modified_base_calls.txt by matching with the sequencing_summary.txt, but I find the problem of transforming theper_read_modified_base_calls.txt to bed. Is there any way of converting a filtered file back to bed? I've tried megalodon_extras aggregate run but does not seem to be intended to aggregate per_read_modified_base_calls.txt to modified_bases.5hmC.bed and modified_bases.5mC.bed, neither does megalodon_extras modified_bases create_motif_bed.

3. Other option I see from the comments is using the modbam file, which I did not output in my previous megalodon runs. Can you get mod_bam from .db files or does megalodon need to be re run?

Finally, as megalodon is being deprecated in favor of Remora and guppy I've been checking the documentation for them, but Remora github documentation seems to rely on megalodon and taiyaki (both deprecated) and guppy documentation does not include much about how Remora can be used to get methylation and hydroxy methylation calls effectively. If we are supposed to change from megalodon to guppy/remora, is there any good documentation on how to actually call modified bases with this new pipeline? I'm also wondering whether the commands --align_ref and --bam_out commands will align all fastq on only pass ones when using them in Guppy or will it behave the same as megalodon.

Sorry for the long question.
Thanks,
Cora