How to extract one modification at a time- used to use --call-code parameter?

[haileye4]

Hey!!

I have an aligned bam from direct RNA sequencing, and it has multiple mods attached to it (m6A, 5mC, 5hmC, and pseudouridine).

Around 6 months ago we ran into this issue where when we would use modkit extract calls to extract multiple modification from RNA data, it would create an insanely large bam and/or crash out on memory and couldn't finish the process because it was using so much RAM (even after throttling with setting threads, queue-size, and interval-size). I solved this problem by using the --call-code parameter and doing one mod at a time, decreasing memory and size of our output files.

Now, I'm noticing --call-code is not even a parameter or option anymore in the latest versions of modkit. I can't seem to figure out 1) what older version I was using 6ish months ago and 2) what versions actually had that parameter as an option.

If I want to use modkit and extract one modification at a time, is there another way to do so rather than just doing a awk command after it is complete? Or, can you tell me what was the last version to use --call-code so I can pull a respective container with the right version?

Lastly, is this something you can add back in? Because at least for us it was extremely useful and our bams were so large we couldn't even get the process to work without doing one modification at a time. Thanks!

Hailey Z

[ArtRand]

Hello @haileye4,

If you use a recent (v0.5.0+) version of Modkit (without --call-code) but add --ignore-index to your command line, do you still see large memory usage? It's true that the extract tables can be quite large, especially for dRNA where you have a modification at every (or nearly every) primary base. If you use --bgzf are the files still unmanageably large? For what it's worth, I'm working on speeding up extract [calls|full] to bring those commands up to date with pileup. If subsetting to specific modifications in order to save space is a useful functionality, I can add something like that back.

[haileye4]

Hey @ArtRand !

I've never tried using --ignore-index before, I'll try it and let you know! I also would do --bgzf but for our analysis, we want to pipe the calls.tsv into another function right after so compressing it won't be helpful.

That's great to know! Thank you! and yes, at least for our purposes subsetting to specific modifications would be a super useful functionality I would love to use!

[ArtRand]

Hello @haileye4,

Sounds good. Let me know if --ignore-index doesn't solve the memory problem and I'll ping you when I have an updated version of extract.