search

microbiome / NMGS

Neutral model
3 stars 7 forks source link

Segmentation Fault #5

Open ramalok opened 7 years ago

ramalok commented 7 years ago

Hi, I'm running NMGS with microbial datasets. It runs OK with one containing 7000 OTUs and 122 samples using 20000 iterations (burnin 10000). Then, when I try another dataset with 122 samples but 17000 OTUs I get a Segmentation Fault error, even though there's plenty of RAM. I've tried reducing the number of iterations to see if that's the problem but same error pops up. Now I'm running it with more iterations to see what happens. Any idea/suggestion? I've compiled it Ubuntu, and there are a couple of warnings during compilation, but that doesn't seem to affect the program, as the test run functions OK. Do you have a binary linux version that I could try just to check it is not a compilation issue? thanks in advance, Ramiro

chrisquince commented 7 years ago

Hi Ramiro,

Hope you are well! This may be related to the file size in fact, maybe try:

define MAX_LINE_LENGTH 10000

changing this constant in the NMGS.h header to say 100 times bigger. Otherwise it is an issue with the construction of the Stirling matrix. What you could try for that is subsampling down your data with rarefaction to reduce the size of the problem. It is probably a good idea to always rarefy down to the size of the smallest sample anyway.

Best Wishes, Chris

apascualgarcia commented 7 years ago

Hi Ramiro,

I had the same issue, and in my case it turned out to be just the length of the labels (of the samples in this case) --they were too long--, after reducing the name it works normally.

Best wishes,

Alberto