# Oxford nanopore promethion; does guppy3 separate fast5 into pass and fail? If not what does?

just a quick question about fast5 pass fail folders. I have been working under the assumption that the fast5 pass fail folders I was given as raw data from our ONT vendor came from them doing base calling with guppy 3.something. Is that right? (The person that worked for the vendor has since moved on so I can't ask them - so my question is really does guppy3 divide fast5s into pass/fail?)

I further assumed that given I am now doing my own basecalling with guppy4.4 that I should use all the fast5s (both pass and fail) as some of the fails might become pass with the improved NN. Again, am I right? If not could someone please explain?

I'm going to also put this question on the ONT community site, FYI

Thanks,

Liam

## 1 Answer

Only if they've enabled the --qscore_filtering flag when using Guppy; this flag isn't enabled by default when running Guppy from the command line.

It's a good idea to use all fast5 reads for recalling, because accuracy improves over time, so some that previously failed may slide into the pass folder.

FYI, I'd be much happier if people asked their nanopore sequencing questions here. It's a publicly-accessible website which is indexed by search engines, so makes it easier to find answers, and easier to see the answers when found (i.e. no login needed).

Note that when recalling, you set the input and output directories (I think via the -i and -s flags). These should be different directories, and ideally different output directories for each recalling attempt, to avoid duplicate file entries. The original fast5 files shouldn't be modified by recalling as long as the input and output directories are different.

• Thanks. The issue I face is that I have now got fast5s of the same name in both pass and fail folders. I want to keep splitting fast5s on qscore. I'm concerned then when I do basecalling I will over write fast5s output. Mar 4 '21 at 6:36