search

jyyulab / SJARACNe

Scalable Tool for Gene Network Reverse Engineering
Other
22 stars 16 forks source link

Did not find output with glob pattern : ['TF_run_001.adj'] #52

Open alistairapoll opened 1 month ago

alistairapoll commented 1 month ago

I am running SJARACNe through a VM using Oracle VM Virtual Box using a Ubuntu distro (24.04.01). I am running into the following error:

Did not find output with glob pattern : ['TF_run_001.adj']

This appears for every iteration of the bootstrapping. This appears to be because there are p values returning as 0 and p is restricted to (0,1], and thus I assume it is not creating the tmp TF_run_00X.adj file in the first place? I have tried uninstalling and reinstalling as others have suggested, but this does not resolve the issue.

image

I have also noticed that it throws the following error earlier in the output:

image

Originally, this CUDA error said it could not find nvidia-smi file, but after reinstalling the nvidia driver, it still throws the non-zero exit status above. However, as it continues to run after this I don't know whether this error results in the downstream issue with the p-values and TF_run_00X.adj files, or whether they are separate issues.

Does anyone have any idea how to overcome this ?

Thanks, Alistair

karlie002 commented 1 month ago

Hi ! I met the same problem just now !!! Does anyone know the solution ? Thanks a lot !

alistairapoll commented 1 month ago

Hi ! I met the same problem just now !!! Does anyone know the solution ? Thanks a lot !

I have found something! The -pb argument sets the bootstrap p value (as opposed to -pc which sets the consensus value) - this should default to 1e-7 but it appears as thought it's defaulting to p=0... So setting -pb 1e-5 has allowed it to run. Hope that helps you too!

karlie002 commented 1 month ago

Hi Alistair ! Thanks a lot for your kind reply , I just tried as you suggested and this error disappeared .

Hi ! I met the same problem just now !!! Does anyone know the solution ? Thanks a lot !

I have found something! The -pb argument sets the bootstrap p value (as opposed to -pc which sets the consensus value) - this should default to 1e-7 but it appears as thought it's defaulting to p=0... So setting -pb 1e-5 has allowed it to run. Hope that helps you too!