I feel like that question is a trap ...

How do you read from the wiff directly? (I am not super familiar with the state of SDKs for wiff/wiff2, thanks beforehand)

I think @ypriverol.bsky.social might have some hard data on this one

In the past I have written stuff to delete things more than X years old that are not raw file and that seems to already give a pretty good compromise. pdresults, pep.xmls, maxquant .peaks and that kind of stuff are massive files. (If I recall my past life that was taking up ~75% of the space)

Well that is false ... I calculated it for 120 TB, not 380 ... so it would actually be ~ 475 USD/month

Couldnt help myself from crunching the numbers of how much this would be in the cloud ... turns out its ~ 155 bucks/month cloud.google.com/products/cal...

It also depends on the tool/acquisition method. Some can be understood as missing at random and some cannot (or at least different ratios of the two)

bsky is descending into degeneracy D:
bsky.app/profile/ucdp...

@swillems.bsky.social do you have any insights on this one ? From the data I've seen the pro data is a bit larger than the ultra series ... Maybe you used compression on the pro and not the ultra ?

Just fyi ... Lfq is something that is not supported on the releases yet but we are thinking on how to have a good implementation for it. (We have an experimental implementation .. DM me if you want to try it out)

Why did the project name have a space ? What kind of savagery is that? :P

I think it is very interesting but I was wondering why your approach does not deal explicitly with missing values (more accurately, missing values are excluded from the CV calculation). Is the assumption that all missing values are missing at random here?

Some of us like writting the software a lot more than any of those :P