https://nf-co.re logo
Join Slack
Powered by
Going through issues now, do we not have the expor...
# drop
a
Going through issues now, do we not have the exporting counts scripts ported over?
n
Hmmm I'm not sure actually, which ones would that be?
a
n
Good catch, I'll try and do it tomorrow
🙌🏼 1
I have a small issue in the aberrant splicing export script: https://github.com/gagneurlab/drop/blob/90fa2b6a43e03f5cd124c1b734be15bede1385dc/drop/modules/aberrant-splicing-pipeline/Counting/exportCounts.R#L47 This line causes the following error in some test data:
Copy code
Error in if (fitMetrics(fds) == "jaccard") { : 
      the condition has length > 1
When I print out 
fitMetrics(fds)
I get the following output:
Copy code
Intron Jaccard Index                 psi5                 psi3 
               "jaccard"               "psi5"               "psi3" 
                   theta 
                 "theta"
Can I update the script to check if this contains "jaccard" instead or how should I handle this situation so that I don't break anything?
a
That is interesting. Which FRASER version is this happening in? FRASER1? psyTypes should be set here I think: https://github.com/nf-core/drop/blob/11452b2e3570868ad988800e883f09afd9778d5e/assets/helpers/aberrant_splicing_config.R#L67 and your output is completely different.
n
It uses 
FRASER2
I'll double check that everything is configured correctly
Okay so apparently setting those options via the config in a function doesn't work. I've added the logic to the script itself and double checked that that code is actually being run, but it's still failing with that error
Copy code
if("$fraser_version" == "FRASER2"){
    pseudocount(0.1)
    psiTypes <- c("jaccard")
    psiTypesNotUsed <- c("psi5", "psi3", "theta")
} else{
    pseudocount(1)
    psiTypes <- c("psi5", "psi3", "theta")
    psiTypesNotUsed <- c("jaccard")
}
I have this now in the script
a
are you now checking the fitmetrics of fds or psiTypes in the final if?
n
fitMetrics still, I haven't changed the if logic because I don't know the background of that part of the code
a
also if it doesnt work here, is there a chance that is doesn't work in the other scripts? 😅
n
I'll try it out using a fresh run, maybe the test data is a bit outdated
a
The logic should be like this:
Copy code
fitMetrics(fds) <- psiTypes
but I'm not sure, maybe it is not working as expected. I think we have to call the configure_fraser first here: https://github.com/nf-core/drop/blob/11452b2e3570868ad988800e883f09afd9778d5e/modu[…]fraser/fithyperparams/templates/04_fit_hyperparameters_FraseR.R before setting the psiTypes.
nvm we do: configure_fraser("$fraser_version")
👍 1
n
exportCounts is using the 
psiValueCalculation
output, right? it hasn't run the fit hyperparameters by then
👍🏼 1
I got it running with a full run on the test data starting from BAM files, but still the same error
I found the issue! the psiTypes weren't correctly set in the PSI value calculation script
👀 1
a
n
correct, I'll add the fix to my upcoming PR
Just running a test now to be 100% sure
a
Sounds good, thanks!
🎉 1
n
Yes, it works now
🔥 1
a
Hi @Nicolas Vannieuwkerke, I was on vacation past week and back today. I'm going through a backlog of things that I have to do. I will check the pr latest Friday.
n
Hi, no worries! Welcome back 😛
🙌🏼 1
Open in Slack
PreviousNext
Powered by