mateidavid / fast5

A C++ header-only library for reading Oxford Nanopore Fast5 files
MIT License
53 stars 16 forks source link

f5pack compession ratio is only 30-40% with r9.4 / albacore2 #21

Closed lpryszcz closed 5 years ago

lpryszcz commented 5 years ago

Hi, in the blog it's mentioned f5pack does 10x compression. But on our runs (r9.4, albacore 2) we only achieve ~40% compression. Is it expected? Are there any other ways of improving fast5 compression? ~/src/fast5/python/bin/f5pack --archive -R -o f5pack/$d reads/$d

jts commented 5 years ago

When f5pack and that blog post was written the fast5 files contained much more information (events annotated by the basecaller). This is no longer the case so the compression ratio will be lower.

Note f5pack is an exploratory tool that we don’t support, so we don’t suggest running it in production.

Jared

On Mar 16, 2019, at 12:24 PM, Leszek Pryszcz notifications@github.com wrote:

Hi, in the blog it's mentioned f5pack does 10x compression. But on our runs (r9.4, albacore 2) we only achieve 30-40% compression. Is it expected? Are there any other ways of improving fast5 compression? ~/src/fast5/python/bin/f5pack --archive -R -o f5pack/$d reads/$d

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

lpryszcz commented 5 years ago

Thanks for clarification!