Arthur Rand

Hello @qiangfu-eu so I've spent some time working on this. I was able to reproduce the results you've shown on the pUC19 reads with modkit and with pysam. Both modkit...

@qiangfu-eu As far as the difference you're seeing in the human data, it could be a due to one or a few things. Firstly, you may be somehow seeing the...

@qiangfu-eu Sorry about the delay in getting back to you. I haven't updated the pileup engine yet, so I wouldn't expect the results would change with the more recent releases....

@qiangfu-eu yes I can check as to the differences in `no_call` counts. Thanks for digging into this! Would you be able to share the BAM you're using for this test?...

Secondary, duplicate, and unmapped reads _shouldn't_ have any effect, thanks for the BAMs, I'll take a look at these and get back.

@qiangfu-eu I think I've gotten to the bottom of this (or at least discovered the source of the difference). `modbam2bed` uses `INT_MAX` ([ref](https://cplusplus.com/reference/climits/)) for the default max depth setting to...

@qiangfu-eu Sorry, could you show me an example (result from the `awk` command) of where the 5mC/5hmC calls are the same but the canonical calls are different? Or are you...

@qiangfu-eu release [0.1.11](https://github.com/nanoporetech/modkit/releases/tag/v0.1.11) has the `--max-depth` parameter ([docs](https://nanoporetech.github.io/modkit/advanced_usage.html#pileup)).

@qiangfu-eu I have not forgotten about this issue, just been busy with some other stuff. I should be able to get to a conclusion this week.

Hello @Wensu-Liu, The differences in modbam2bed compared to modkit have to do with the [hierarchical threshold algorithm](https://nanoporetech.github.io/modkit/filtering_details.html) used in modkit. We've seen that this thresholding mechanism produces the best results....