Closed ireneg closed 7 years ago
The negative length vectors not allowed
error message means your R session has run out of memory. You can either try to run the simulation on a machine with more memory, or you could break up the input files into smaller input files and do partial simulations; then concatenate the output.
As for the replicates question, it looks like you're using the current release version (1.8.3), which per the docs on num_reps
is the number of replicates in each group -- so if it's a single integer, you're simulating 2 groups of 10 replicates each (20 replicates).
That said, the behavior should be that the length of the vector determines the number of groups, as of a refactor I made in Dec 2014 :/ I will check on why the release version is so outdated. In the meantime, you can install the github version using the install_github function from devtools: install_github('alyssafrazee/polyester')
.
Thank you so much on both counts! I hadn't even thought about memory but our cluster is very weird in how it assigns it, so I might not be getting nearly as much as I thought, which would explain things, especially since the files that work are subsets of the ones that fail.
And yes, I had looked at your example code here on github and it seemed that you specified two groups explicitly, so that also didn't occur to me. I was just wondering where my extra factor of 2 was coming from!
Thanks again! Back to running all my other simulations now :)
Hi guys,
I'm trying to simulate sequence under a bunch of different conditions. The vast majority of them work, but I've been hitting an uninformative error message with others:
Often times,
simulate_experiment
fails fairly quickly with this error message:I launch many jobs with the same script, for multiple values of extraVars[2], and only some are repeatedly and predictably failing; extraVars[3] is always either 50 or 100. I realise the problem is probably coming from my fasta input, but the files I use differ by thousands of lines and checking them manually is not really going to work, so any intuition you might have as to what is causing the error would really help me clean them up - thank you!!
(And, not to be captain obvious here, but here are the lengths for one of the failed runs:
there are no obvious negative lengths that I can find.)
Oh, also - for some reason this command yields 20 replicates instead of 10 per run of simulate_experiment. Any idea why? Not that I don't appreciate having twice as many samples as I need, I just couldn't find the answer in the manual.